bibtex_url
stringlengths 41
53
| proceedings
stringlengths 38
50
| bibtext
stringlengths 528
3.02k
| abstract
stringlengths 17
2.35k
| authors
sequencelengths 1
44
| title
stringlengths 18
190
| id
stringlengths 7
19
| arxiv_id
stringlengths 0
10
| GitHub
sequencelengths 1
1
| paper_page
stringclasses 528
values | n_linked_authors
int64 -1
15
| upvotes
int64 -1
77
| num_comments
int64 -1
10
| n_authors
int64 -1
52
| Models
sequencelengths 0
100
| Datasets
sequencelengths 0
15
| Spaces
sequencelengths 0
46
| paper_page_exists_pre_conf
int64 0
1
| type
stringclasses 2
values |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
https://aclanthology.org/2023.findings-emnlp.520.bib | https://aclanthology.org/2023.findings-emnlp.520/ | @inproceedings{aida-bollegala-2023-swap,
title = "Swap and Predict {--} Predicting the Semantic Changes in Words across Corpora by Context Swapping",
author = "Aida, Taichi and
Bollegala, Danushka",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.520",
doi = "10.18653/v1/2023.findings-emnlp.520",
pages = "7753--7772",
abstract = "Meanings of words change over time and across domains. Detecting the semantic changes of words is an important task for various NLP applications that must make time-sensitive predictions. We consider the problem of predicting whether a given target word, $w$, changes its meaning between two different text corpora, $\mathcal{C}_1$ and $\mathcal{C}_2$. For this purpose, we propose $\textit{Swapping-based Semantic Change Detection}$ (SSCD), an unsupervised method that randomly swaps contexts between $\mathcal{C}_1$ and $\mathcal{C}_2$ where $w$ occurs. We then look at the distribution of contextualised word embeddings of $w$, obtained from a pretrained masked language model (MLM), representing the meaning of $w$ in its occurrence contexts in $\mathcal{C}_1$ and $\mathcal{C}_2$. Intuitively, if the meaning of $w$ does not change between $\mathcal{C}_1$ and $\mathcal{C}_2$, we would expect the distributions of contextualised word embeddings of $w$ to remain the same before and after this random swapping process. Despite its simplicity, we demonstrate that even by using pretrained MLMs without any fine-tuning, our proposed context swapping method accurately predicts the semantic changes of words in four languages (English, German, Swedish, and Latin) and across different time spans (over 50 years and about five years). Moreover, our method achieves significant performance improvements compared to strong baselines for the English semantic change prediction task. Source code is available at https://github.com/a1da4/svp-swap .",
}
| Meanings of words change over time and across domains. Detecting the semantic changes of words is an important task for various NLP applications that must make time-sensitive predictions. We consider the problem of predicting whether a given target word, $w$, changes its meaning between two different text corpora, $\mathcal{C}_1$ and $\mathcal{C}_2$. For this purpose, we propose $\textit{Swapping-based Semantic Change Detection}$ (SSCD), an unsupervised method that randomly swaps contexts between $\mathcal{C}_1$ and $\mathcal{C}_2$ where $w$ occurs. We then look at the distribution of contextualised word embeddings of $w$, obtained from a pretrained masked language model (MLM), representing the meaning of $w$ in its occurrence contexts in $\mathcal{C}_1$ and $\mathcal{C}_2$. Intuitively, if the meaning of $w$ does not change between $\mathcal{C}_1$ and $\mathcal{C}_2$, we would expect the distributions of contextualised word embeddings of $w$ to remain the same before and after this random swapping process. Despite its simplicity, we demonstrate that even by using pretrained MLMs without any fine-tuning, our proposed context swapping method accurately predicts the semantic changes of words in four languages (English, German, Swedish, and Latin) and across different time spans (over 50 years and about five years). Moreover, our method achieves significant performance improvements compared to strong baselines for the English semantic change prediction task. Source code is available at https://github.com/a1da4/svp-swap . | [
"Aida, Taichi",
"Bollegala, Danushka"
] | Swap and Predict – Predicting the Semantic Changes in Words across Corpora by Context Swapping | findings-emnlp.520 | [
"https://github.com/a1da4/svp-swap"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.521.bib | https://aclanthology.org/2023.findings-emnlp.521/ | @inproceedings{zhu-etal-2023-beyond-layout,
title = "Beyond Layout Embedding: Layout Attention with {G}aussian Biases for Structured Document Understanding",
author = "Zhu, Xi and
Han, Xue and
Peng, Shuyuan and
Lei, Shuo and
Deng, Chao and
Feng, Junlan",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.521",
doi = "10.18653/v1/2023.findings-emnlp.521",
pages = "7773--7784",
abstract = "Effectively encoding layout information is a central problem in structured document understanding. Most existing methods rely heavily on millions of trainable parameters to learn the layout features of each word from Cartesian coordinates. However, two unresolved questions remain: (1) Is the Cartesian coordinate system the optimal choice for layout modeling? (2) Are massive learnable parameters truly necessary for layout representation? In this paper, we address these questions by proposing Layout Attention with Gaussian Biases (LAGaBi): Firstly, we find that polar coordinates provide a superior choice over Cartesian coordinates as they offer a measurement of both distance and angle between word pairs, capturing relative positions more effectively. Furthermore, by feeding the distances and angles into 2-D Gaussian kernels, we model intuitive inductive layout biases, i.e., the words closer within a document should receive more attention, which will act as the attention biases to revise the textual attention distribution. LAGaBi is model-agnostic and language-independent, which can be applied to a range of transformer-based models, such as the text pre-training models from the BERT series and the LayoutLM series that incorporate visual features. Experimental results on three widely used benchmarks demonstrate that, despite reducing the number of layout parameters from millions to 48, LAGaBi achieves competitive or even superior performance.",
}
| Effectively encoding layout information is a central problem in structured document understanding. Most existing methods rely heavily on millions of trainable parameters to learn the layout features of each word from Cartesian coordinates. However, two unresolved questions remain: (1) Is the Cartesian coordinate system the optimal choice for layout modeling? (2) Are massive learnable parameters truly necessary for layout representation? In this paper, we address these questions by proposing Layout Attention with Gaussian Biases (LAGaBi): Firstly, we find that polar coordinates provide a superior choice over Cartesian coordinates as they offer a measurement of both distance and angle between word pairs, capturing relative positions more effectively. Furthermore, by feeding the distances and angles into 2-D Gaussian kernels, we model intuitive inductive layout biases, i.e., the words closer within a document should receive more attention, which will act as the attention biases to revise the textual attention distribution. LAGaBi is model-agnostic and language-independent, which can be applied to a range of transformer-based models, such as the text pre-training models from the BERT series and the LayoutLM series that incorporate visual features. Experimental results on three widely used benchmarks demonstrate that, despite reducing the number of layout parameters from millions to 48, LAGaBi achieves competitive or even superior performance. | [
"Zhu, Xi",
"Han, Xue",
"Peng, Shuyuan",
"Lei, Shuo",
"Deng, Chao",
"Feng, Junlan"
] | Beyond Layout Embedding: Layout Attention with Gaussian Biases for Structured Document Understanding | findings-emnlp.521 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.522.bib | https://aclanthology.org/2023.findings-emnlp.522/ | @inproceedings{li-etal-2023-espvr,
title = "{ESPVR}: Entity Spans Position Visual Regions for Multimodal Named Entity Recognition",
author = "Li, Xiujiao and
Sun, Guanglu and
Liu, Xinyu",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.522",
doi = "10.18653/v1/2023.findings-emnlp.522",
pages = "7785--7794",
abstract = "Multimodal Named Entity Recognition (MNER) uses visual information to improve the performance of text-only Named Entity Recognition (NER). However, existing methods for acquiring local visual information suffer from certain limitations: (1) using an attention-based method to extract visual regions related to the text from visual regions obtained through convolutional architectures (e.g., ResNet), attention is distracted by the entire image, rather than being fully focused on the visual regions most relevant to the text; (2) using an object detection-based (e.g., Mask R-CNN) method to detect visual object regions related to the text, object detection has a limited range of recognition categories. Moreover, the visual regions obtained by object detection may not correspond to the entities in the text. In summary, the goal of these methods is not to extract the most relevant visual regions for the entities in the text. The visual regions obtained by these methods may be redundant or insufficient for the entities in the text. In this paper, we propose an Entity Spans Position Visual Regions (ESPVR) module to obtain the most relevant visual regions corresponding to the entities in the text. Experiments show that our proposed approach can achieve the SOTA on Twitter-2017 and competitive results on Twitter-2015.",
}
| Multimodal Named Entity Recognition (MNER) uses visual information to improve the performance of text-only Named Entity Recognition (NER). However, existing methods for acquiring local visual information suffer from certain limitations: (1) using an attention-based method to extract visual regions related to the text from visual regions obtained through convolutional architectures (e.g., ResNet), attention is distracted by the entire image, rather than being fully focused on the visual regions most relevant to the text; (2) using an object detection-based (e.g., Mask R-CNN) method to detect visual object regions related to the text, object detection has a limited range of recognition categories. Moreover, the visual regions obtained by object detection may not correspond to the entities in the text. In summary, the goal of these methods is not to extract the most relevant visual regions for the entities in the text. The visual regions obtained by these methods may be redundant or insufficient for the entities in the text. In this paper, we propose an Entity Spans Position Visual Regions (ESPVR) module to obtain the most relevant visual regions corresponding to the entities in the text. Experiments show that our proposed approach can achieve the SOTA on Twitter-2017 and competitive results on Twitter-2015. | [
"Li, Xiujiao",
"Sun, Guanglu",
"Liu, Xinyu"
] | ESPVR: Entity Spans Position Visual Regions for Multimodal Named Entity Recognition | findings-emnlp.522 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.523.bib | https://aclanthology.org/2023.findings-emnlp.523/ | @inproceedings{shen-etal-2023-flatness,
title = "Flatness-Aware Prompt Selection Improves Accuracy and Sample Efficiency",
author = "Shen, Lingfeng and
Tan, Weiting and
Zheng, Boyuan and
Khashabi, Daniel",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.523",
doi = "10.18653/v1/2023.findings-emnlp.523",
pages = "7795--7817",
abstract = "With growing capabilities of large language models, prompting them has become the dominant way to access them. This has motivated the development of strategies for automatically selecting effective language prompts. In this paper, we introduce **pFlat** (prompt flatness), a new metric to quantify the expected utility of a language prompt. This metric is inspired by *flatness* regularization in statistical learning that quantifies the robustness of the model towards its parameter perturbations. We provide theoretical foundations for this metric and its relationship with other prompt selection metrics, providing a comprehensive understanding of existing methods. Empirically, we show that combining **pFlat** with existing metrics improves both performance and sample efficiency. Our metric outperforms the previous prompt selection metrics with an average increase of 10{\%} in Pearson correlation across 6 classification benchmarks, and the prompt selected by our metric gains 5{\%} higher accuracy than previous metrics across the benchmarks.",
}
| With growing capabilities of large language models, prompting them has become the dominant way to access them. This has motivated the development of strategies for automatically selecting effective language prompts. In this paper, we introduce **pFlat** (prompt flatness), a new metric to quantify the expected utility of a language prompt. This metric is inspired by *flatness* regularization in statistical learning that quantifies the robustness of the model towards its parameter perturbations. We provide theoretical foundations for this metric and its relationship with other prompt selection metrics, providing a comprehensive understanding of existing methods. Empirically, we show that combining **pFlat** with existing metrics improves both performance and sample efficiency. Our metric outperforms the previous prompt selection metrics with an average increase of 10{\%} in Pearson correlation across 6 classification benchmarks, and the prompt selected by our metric gains 5{\%} higher accuracy than previous metrics across the benchmarks. | [
"Shen, Lingfeng",
"Tan, Weiting",
"Zheng, Boyuan",
"Khashabi, Daniel"
] | Flatness-Aware Prompt Selection Improves Accuracy and Sample Efficiency | findings-emnlp.523 | 2305.10713 | [
"https://github.com/shadowkiller33/flatness"
] | https://huggingface.co/papers/2305.10713 | 1 | 0 | 0 | 4 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.524.bib | https://aclanthology.org/2023.findings-emnlp.524/ | @inproceedings{pavlopoulos-etal-2023-detecting,
title = "Detecting Erroneously Recognized Handwritten Byzantine Text",
author = "Pavlopoulos, John and
Kougia, Vasiliki and
Platanou, Paraskevi and
Essler, Holger",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.524",
doi = "10.18653/v1/2023.findings-emnlp.524",
pages = "7818--7828",
abstract = "Handwritten text recognition (HTR) yields textual output that comprises errors, which are considerably more compared to that of recognised printed (OCRed) text. Post-correcting methods can eliminate such errors but may also introduce errors. In this study, we investigate the issues arising from this reality in Byzantine Greek. We investigate the properties of the texts that lead post-correction systems to this adversarial behaviour and we experiment with text classification systems that learn to detect incorrect recognition output. A large masked language model, pre-trained in modern and fine-tuned in Byzantine Greek, achieves an Average Precision score of 95{\%}. The score improves to 97{\%} when using a model that is pre-trained in modern and then in ancient Greek, the two language forms Byzantine Greek combines elements from. A century-based analysis shows that the advantage of the classifier that is further-pre-trained in ancient Greek concerns texts of older centuries. The application of this classifier before a neural post-corrector on HTRed text reduced significantly the post-correction mistakes.",
}
| Handwritten text recognition (HTR) yields textual output that comprises errors, which are considerably more compared to that of recognised printed (OCRed) text. Post-correcting methods can eliminate such errors but may also introduce errors. In this study, we investigate the issues arising from this reality in Byzantine Greek. We investigate the properties of the texts that lead post-correction systems to this adversarial behaviour and we experiment with text classification systems that learn to detect incorrect recognition output. A large masked language model, pre-trained in modern and fine-tuned in Byzantine Greek, achieves an Average Precision score of 95{\%}. The score improves to 97{\%} when using a model that is pre-trained in modern and then in ancient Greek, the two language forms Byzantine Greek combines elements from. A century-based analysis shows that the advantage of the classifier that is further-pre-trained in ancient Greek concerns texts of older centuries. The application of this classifier before a neural post-corrector on HTRed text reduced significantly the post-correction mistakes. | [
"Pavlopoulos, John",
"Kougia, Vasiliki",
"Platanou, Paraskevi",
"Essler, Holger"
] | Detecting Erroneously Recognized Handwritten Byzantine Text | findings-emnlp.524 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.525.bib | https://aclanthology.org/2023.findings-emnlp.525/ | @inproceedings{xue-etal-2023-improving,
title = "Improving Factual Consistency for Knowledge-Grounded Dialogue Systems via Knowledge Enhancement and Alignment",
author = "Xue, Boyang and
Wang, Weichao and
Wang, Hongru and
Mi, Fei and
Wang, Rui and
Wang, Yasheng and
Shang, Lifeng and
Jiang, Xin and
Liu, Qun and
Wong, Kam-Fai",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.525",
doi = "10.18653/v1/2023.findings-emnlp.525",
pages = "7829--7844",
abstract = "Pretrained language models (PLMs) based knowledge-grounded dialogue systems are prone to generate responses that are factually inconsistent with the provided knowledge source. In such inconsistent responses, the dialogue models fail to accurately express the external factual knowledge they rely upon. Inspired by previous work which identified that feedforward networks (FFNs) within Transformers are responsible for factual knowledge expressions, we investigate two methods to efficiently improve the factual expression capability of FFNs by knowledge enhancement and alignment respectively. We first propose K-Dial, which explicitly introduces extended FFNs in Transformers to enhance factual knowledge expressions given the specific patterns of knowledge-grounded dialogue inputs. Additionally, we apply the reinforcement learning for factual consistency (RLFC) method to implicitly adjust FFNs{'} expressions in responses by aligning with gold knowledge for the factual consistency preference. To comprehensively assess the factual consistency and dialogue quality of responses, we employ extensive automatic measures and human evaluations including sophisticated fine-grained NLI-based metrics. Experimental results on WoW and CMU{\_}DoG datasets demonstrate that our methods efficiently enhance the ability of the FFN module to convey factual knowledge, validating the efficacy of improving factual consistency for knowledge-grounded dialogue systems.",
}
| Pretrained language models (PLMs) based knowledge-grounded dialogue systems are prone to generate responses that are factually inconsistent with the provided knowledge source. In such inconsistent responses, the dialogue models fail to accurately express the external factual knowledge they rely upon. Inspired by previous work which identified that feedforward networks (FFNs) within Transformers are responsible for factual knowledge expressions, we investigate two methods to efficiently improve the factual expression capability of FFNs by knowledge enhancement and alignment respectively. We first propose K-Dial, which explicitly introduces extended FFNs in Transformers to enhance factual knowledge expressions given the specific patterns of knowledge-grounded dialogue inputs. Additionally, we apply the reinforcement learning for factual consistency (RLFC) method to implicitly adjust FFNs{'} expressions in responses by aligning with gold knowledge for the factual consistency preference. To comprehensively assess the factual consistency and dialogue quality of responses, we employ extensive automatic measures and human evaluations including sophisticated fine-grained NLI-based metrics. Experimental results on WoW and CMU{\_}DoG datasets demonstrate that our methods efficiently enhance the ability of the FFN module to convey factual knowledge, validating the efficacy of improving factual consistency for knowledge-grounded dialogue systems. | [
"Xue, Boyang",
"Wang, Weichao",
"Wang, Hongru",
"Mi, Fei",
"Wang, Rui",
"Wang, Yasheng",
"Shang, Lifeng",
"Jiang, Xin",
"Liu, Qun",
"Wong, Kam-Fai"
] | Improving Factual Consistency for Knowledge-Grounded Dialogue Systems via Knowledge Enhancement and Alignment | findings-emnlp.525 | 2310.08372 | [
"https://github.com/amourwaltz/factdial"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.526.bib | https://aclanthology.org/2023.findings-emnlp.526/ | @inproceedings{lu-etal-2023-trip,
title = "{TRIP}: Accelerating Document-level Multilingual Pre-training via Triangular Document-level Pre-training on Parallel Data Triplets",
author = "Lu, Hongyuan and
Huang, Haoyang and
Ma, Shuming and
Zhang, Dongdong and
Lam, Wai and
Gao, Zhaochuan and
Aue, Anthony and
Menezes, Arul and
Wei, Furu",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.526",
doi = "10.18653/v1/2023.findings-emnlp.526",
pages = "7845--7858",
abstract = "Despite the success of multilingual sequence-to-sequence pre-training, most existing approaches rely on document-level monolingual corpora in many different languages, sentence-level bilingual corpora, and sometimes synthetic document-level bilingual corpora. This hampers the performance with cross-lingual document-level tasks such as document-level translation. Hence, we propose to mine and leverage document-level trilingual parallel corpora to improve sequence-to-sequence multilingual pre-training. We present \textbf{Tri}angular Document-level \textbf{P}re-training (\textbf{TRIP}) as the first in the field to accelerate the conventional monolingual and bilingual objectives into a trilingual objective with a novel method called Grafting. Experiments show that TRIP achieves several strong state-of-the-art (SOTA) scores on three multilingual document-level machine translation benchmarks and one cross-lingual abstractive summarization benchmark, including consistent improvements by up to 3.11 d-BLEU points and 8.9 ROUGE-L points.",
}
| Despite the success of multilingual sequence-to-sequence pre-training, most existing approaches rely on document-level monolingual corpora in many different languages, sentence-level bilingual corpora, and sometimes synthetic document-level bilingual corpora. This hampers the performance with cross-lingual document-level tasks such as document-level translation. Hence, we propose to mine and leverage document-level trilingual parallel corpora to improve sequence-to-sequence multilingual pre-training. We present \textbf{Tri}angular Document-level \textbf{P}re-training (\textbf{TRIP}) as the first in the field to accelerate the conventional monolingual and bilingual objectives into a trilingual objective with a novel method called Grafting. Experiments show that TRIP achieves several strong state-of-the-art (SOTA) scores on three multilingual document-level machine translation benchmarks and one cross-lingual abstractive summarization benchmark, including consistent improvements by up to 3.11 d-BLEU points and 8.9 ROUGE-L points. | [
"Lu, Hongyuan",
"Huang, Haoyang",
"Ma, Shuming",
"Zhang, Dongdong",
"Lam, Wai",
"Gao, Zhaochuan",
"Aue, Anthony",
"Menezes, Arul",
"Wei, Furu"
] | TRIP: Accelerating Document-level Multilingual Pre-training via Triangular Document-level Pre-training on Parallel Data Triplets | findings-emnlp.526 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.527.bib | https://aclanthology.org/2023.findings-emnlp.527/ | @inproceedings{zevallos-etal-2023-frequency,
title = "Frequency Balanced Datasets Lead to Better Language Models",
author = "Zevallos, Rodolfo and
Farr{\'u}s, Mireia and
Bel, N{\'u}ria",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.527",
doi = "10.18653/v1/2023.findings-emnlp.527",
pages = "7859--7872",
abstract = "This paper reports on the experiments aimed to improve our understanding of the role of the amount of data required for training attention-based transformer language models. Specifically, we investigate the impact of reducing the immense amounts of required pre-training data through sampling strategies that identify and reduce high-frequency tokens as different studies have indicated that the existence of very high-frequency tokens in pre-training data might bias learning, causing undesired effects. In this light, we describe our sampling algorithm that iteratively assesses token frequencies and removes sentences that contain still high-frequency tokens, eventually delivering a balanced, linguistically correct dataset. We evaluate the results in terms of model perplexity and fine-tuning linguistic probing tasks, NLP downstream tasks as well as more semantic SuperGlue tasks. The results show that pre-training with the resulting balanced dataset allows reducing up to three times the pre-training data.",
}
| This paper reports on the experiments aimed to improve our understanding of the role of the amount of data required for training attention-based transformer language models. Specifically, we investigate the impact of reducing the immense amounts of required pre-training data through sampling strategies that identify and reduce high-frequency tokens as different studies have indicated that the existence of very high-frequency tokens in pre-training data might bias learning, causing undesired effects. In this light, we describe our sampling algorithm that iteratively assesses token frequencies and removes sentences that contain still high-frequency tokens, eventually delivering a balanced, linguistically correct dataset. We evaluate the results in terms of model perplexity and fine-tuning linguistic probing tasks, NLP downstream tasks as well as more semantic SuperGlue tasks. The results show that pre-training with the resulting balanced dataset allows reducing up to three times the pre-training data. | [
"Zevallos, Rodolfo",
"Farr{\\'u}s, Mireia",
"Bel, N{\\'u}ria"
] | Frequency Balanced Datasets Lead to Better Language Models | findings-emnlp.527 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.528.bib | https://aclanthology.org/2023.findings-emnlp.528/ | @inproceedings{wang-etal-2023-uncertainty,
title = "Uncertainty-aware Parameter-Efficient Self-training for Semi-supervised Language Understanding",
author = "Wang, Jianing and
Sun, Qiushi and
Chen, Nuo and
Wang, Chengyu and
Huang, Jun and
Gao, Ming and
Li, Xiang",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.528",
doi = "10.18653/v1/2023.findings-emnlp.528",
pages = "7873--7884",
abstract = "The recent success of large pre-trained language models (PLMs) heavily hinges on massive labeled data, which typically produces inferior performance in low-resource scenarios. To remedy this dilemma, we study self-training as one of the predominant semi-supervised learning (SSL) approaches, which utilizes large-scale unlabeled data to generate synthetic examples. However, too many noisy labels will hurt the model performance, and the self-training procedure requires multiple training iterations making it more expensive if all the model parameters of the PLM are updated. This paper presents UPET, a novel Uncertainty-aware Parameter-Efficient self-Training framework to effectively and efficiently address the labeled data scarcity issue. Specifically, we incorporate Monte Carlo (MC) dropout in Bayesian neural network (BNN) to perform uncertainty estimation for the teacher model and then judiciously select reliable pseudo-labeled examples based on confidence and certainty. During the student training, we introduce multiple parameter-efficient learning (PEL) paradigms that allow optimizes only a small percentage of parameters. We also propose a novel Easy-Hard Contrastive Tuning to enhance the robustness and generalization. Extensive experiments over multiple downstream tasks demonstrate that UPET achieves a substantial improvement in terms of performance and efficiency. Our codes and data are released at https: //github.com/wjn1996/UPET.",
}
| The recent success of large pre-trained language models (PLMs) heavily hinges on massive labeled data, which typically produces inferior performance in low-resource scenarios. To remedy this dilemma, we study self-training as one of the predominant semi-supervised learning (SSL) approaches, which utilizes large-scale unlabeled data to generate synthetic examples. However, too many noisy labels will hurt the model performance, and the self-training procedure requires multiple training iterations making it more expensive if all the model parameters of the PLM are updated. This paper presents UPET, a novel Uncertainty-aware Parameter-Efficient self-Training framework to effectively and efficiently address the labeled data scarcity issue. Specifically, we incorporate Monte Carlo (MC) dropout in Bayesian neural network (BNN) to perform uncertainty estimation for the teacher model and then judiciously select reliable pseudo-labeled examples based on confidence and certainty. During the student training, we introduce multiple parameter-efficient learning (PEL) paradigms that allow optimizes only a small percentage of parameters. We also propose a novel Easy-Hard Contrastive Tuning to enhance the robustness and generalization. Extensive experiments over multiple downstream tasks demonstrate that UPET achieves a substantial improvement in terms of performance and efficiency. Our codes and data are released at https: //github.com/wjn1996/UPET. | [
"Wang, Jianing",
"Sun, Qiushi",
"Chen, Nuo",
"Wang, Chengyu",
"Huang, Jun",
"Gao, Ming",
"Li, Xiang"
] | Uncertainty-aware Parameter-Efficient Self-training for Semi-supervised Language Understanding | findings-emnlp.528 | 2310.13022 | [
"https://github.com/wjn1996/upet"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.529.bib | https://aclanthology.org/2023.findings-emnlp.529/ | @inproceedings{li-etal-2023-tr,
title = "{TR}-Rules: Rule-based Model for Link Forecasting on Temporal Knowledge Graph Considering Temporal Redundancy",
author = "Li, Ningyuan and
E, Haihong and
Li, Shi and
Sun, Mingzhi and
Yao, Tianyu and
Song, Meina and
Wang, Yong and
Luo, Haoran",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.529",
doi = "10.18653/v1/2023.findings-emnlp.529",
pages = "7885--7894",
abstract = "Temporal knowledge graph (TKG) has been proved to be an effective way for modeling dynamic facts in real world. Many efforts have been devoted into predicting future events i.e. extrapolation, on TKGs. Recently, rule-based knowledge graph completion methods which are considered to be more interpretable than embedding-based methods, have been transferred to temporal knowledge graph extrapolation. However, rule-based models suffer from temporal redundancy when leveraged under dynamic settings, which results in inaccurate rule confidence calculation. In this paper, we define the problem of temporal redundancy and propose TR-Rules which solves the temporal redundancy issues through a simple but effective strategy. Besides, to capture more information lurking in TKGs, apart from cyclic rules, TR-Rules also mines and properly leverages acyclic rules, which has not been explored by existing models. Experimental results on three benchmarks show that TR-Rules achieves state-of-the-art performance. Ablation study shows the impact of temporal redundancy and demonstrates the performance of acyclic rules is much more promising due to its higher sensitivity to the number of sampled walks during learning stage.",
}
| Temporal knowledge graph (TKG) has been proved to be an effective way for modeling dynamic facts in real world. Many efforts have been devoted into predicting future events i.e. extrapolation, on TKGs. Recently, rule-based knowledge graph completion methods which are considered to be more interpretable than embedding-based methods, have been transferred to temporal knowledge graph extrapolation. However, rule-based models suffer from temporal redundancy when leveraged under dynamic settings, which results in inaccurate rule confidence calculation. In this paper, we define the problem of temporal redundancy and propose TR-Rules which solves the temporal redundancy issues through a simple but effective strategy. Besides, to capture more information lurking in TKGs, apart from cyclic rules, TR-Rules also mines and properly leverages acyclic rules, which has not been explored by existing models. Experimental results on three benchmarks show that TR-Rules achieves state-of-the-art performance. Ablation study shows the impact of temporal redundancy and demonstrates the performance of acyclic rules is much more promising due to its higher sensitivity to the number of sampled walks during learning stage. | [
"Li, Ningyuan",
"E, Haihong",
"Li, Shi",
"Sun, Mingzhi",
"Yao, Tianyu",
"Song, Meina",
"Wang, Yong",
"Luo, Haoran"
] | TR-Rules: Rule-based Model for Link Forecasting on Temporal Knowledge Graph Considering Temporal Redundancy | findings-emnlp.529 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.530.bib | https://aclanthology.org/2023.findings-emnlp.530/ | @inproceedings{zhao-titov-2023-transferability,
title = "On the Transferability of Visually Grounded {PCFG}s",
author = "Zhao, Yanpeng and
Titov, Ivan",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.530",
doi = "10.18653/v1/2023.findings-emnlp.530",
pages = "7895--7910",
abstract = "There has been a significant surge of interest in visually grounded grammar induction in recent times. While a variety of models have been developed for the task and have demonstrated impressive performance, they have not been evaluated on text domains that are different from the training domain, so it is unclear if the improvements brought by visual groundings are transferable. Our study aims to fill this gap and assess the degree of transferability. We start by extending VC-PCFG (short for Visually-grounded Compound PCFG [[Zhao and Titov, 2020](https://aclanthology.org/2020.emnlp-main.354/)]) in such a way that it can transfer across text domains. We consider a zero-shot transfer learning setting where a model is trained on the source domain and is directly applied to target domains, without any further training. Our experimental results suggest that: the benefits from using visual groundings transfer to text in a domain similar to the training domain but fail to transfer to remote domains. Further, we conduct data and result analysis; we find that the lexicon overlap between the source domain and the target domain is the most important factor in the transferability of VC-PCFG.",
}
| There has been a significant surge of interest in visually grounded grammar induction in recent times. While a variety of models have been developed for the task and have demonstrated impressive performance, they have not been evaluated on text domains that are different from the training domain, so it is unclear if the improvements brought by visual groundings are transferable. Our study aims to fill this gap and assess the degree of transferability. We start by extending VC-PCFG (short for Visually-grounded Compound PCFG [[Zhao and Titov, 2020](https://aclanthology.org/2020.emnlp-main.354/)]) in such a way that it can transfer across text domains. We consider a zero-shot transfer learning setting where a model is trained on the source domain and is directly applied to target domains, without any further training. Our experimental results suggest that: the benefits from using visual groundings transfer to text in a domain similar to the training domain but fail to transfer to remote domains. Further, we conduct data and result analysis; we find that the lexicon overlap between the source domain and the target domain is the most important factor in the transferability of VC-PCFG. | [
"Zhao, Yanpeng",
"Titov, Ivan"
] | On the Transferability of Visually Grounded PCFGs | findings-emnlp.530 | 2310.14107 | [
"https://github.com/zhaoyanpeng/cpcfg"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.531.bib | https://aclanthology.org/2023.findings-emnlp.531/ | @inproceedings{kawano-etal-2023-analysis,
title = "Analysis of Style-Shifting on Social Media: Using Neural Language Model Conditioned by Social Meanings",
author = "Kawano, Seiya and
Kanezaki, Shota and
Garcia Contreras, Angel Fernando and
Yuguchi, Akishige and
Katsurai, Marie and
Yoshino, Koichiro",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.531",
doi = "10.18653/v1/2023.findings-emnlp.531",
pages = "7911--7921",
abstract = "In this paper, we propose a novel framework for evaluating style-shifting in social media conversations. Our proposed framework captures changes in an individual{'}s conversational style based on surprisals predicted by a personalized neural language model for individuals. Our personalized language model integrates not only the linguistic contents of conversations but also non-linguistic factors, such as social meanings, including group membership, personal attributes, and individual beliefs. We incorporate these factors directly or implicitly into our model, leveraging large, pre-trained language models and feature vectors derived from a relationship graph on social media. Compared to existing models, our personalized language model demonstrated superior performance in predicting an individual{'}s language in a test set. Furthermore, an analysis of style-shifting utilizing our proposed metric based on our personalized neural language model reveals a correlation between our metric and various conversation factors as well as human evaluation of style-shifting.",
}
| In this paper, we propose a novel framework for evaluating style-shifting in social media conversations. Our proposed framework captures changes in an individual{'}s conversational style based on surprisals predicted by a personalized neural language model for individuals. Our personalized language model integrates not only the linguistic contents of conversations but also non-linguistic factors, such as social meanings, including group membership, personal attributes, and individual beliefs. We incorporate these factors directly or implicitly into our model, leveraging large, pre-trained language models and feature vectors derived from a relationship graph on social media. Compared to existing models, our personalized language model demonstrated superior performance in predicting an individual{'}s language in a test set. Furthermore, an analysis of style-shifting utilizing our proposed metric based on our personalized neural language model reveals a correlation between our metric and various conversation factors as well as human evaluation of style-shifting. | [
"Kawano, Seiya",
"Kanezaki, Shota",
"Garcia Contreras, Angel Fern",
"o",
"Yuguchi, Akishige",
"Katsurai, Marie",
"Yoshino, Koichiro"
] | Analysis of Style-Shifting on Social Media: Using Neural Language Model Conditioned by Social Meanings | findings-emnlp.531 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.532.bib | https://aclanthology.org/2023.findings-emnlp.532/ | @inproceedings{yin-van-schijndel-2023-linguistic,
title = "Linguistic Compression in Single-Sentence Human-Written Summaries",
author = "Yin, Fangcong and
van Schijndel, Marten",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.532",
doi = "10.18653/v1/2023.findings-emnlp.532",
pages = "7922--7935",
abstract = "Summarizing texts involves significant cognitive efforts to compress information. While advances in automatic summarization systems have drawn attention from the NLP and linguistics communities to this topic, there is a lack of computational studies of linguistic patterns in human-written summaries. This work presents a large-scale corpus study of human-written single-sentence summaries. We analyzed the linguistic compression patterns from source documents to summaries at different granularities, and we found that summaries are generally written with morphological expansion, increased lexical diversity, and similar positional arrangements of specific words compared to the source across different genres. We also studied how linguistic compressions of different factors affect reader judgments of quality through a human study, with the results showing that the use of morphological and syntactic changes by summary writers matches reader preferences while lexical diversity and word specificity preferences are not aligned between summary writers and readers.",
}
| Summarizing texts involves significant cognitive efforts to compress information. While advances in automatic summarization systems have drawn attention from the NLP and linguistics communities to this topic, there is a lack of computational studies of linguistic patterns in human-written summaries. This work presents a large-scale corpus study of human-written single-sentence summaries. We analyzed the linguistic compression patterns from source documents to summaries at different granularities, and we found that summaries are generally written with morphological expansion, increased lexical diversity, and similar positional arrangements of specific words compared to the source across different genres. We also studied how linguistic compressions of different factors affect reader judgments of quality through a human study, with the results showing that the use of morphological and syntactic changes by summary writers matches reader preferences while lexical diversity and word specificity preferences are not aligned between summary writers and readers. | [
"Yin, Fangcong",
"van Schijndel, Marten"
] | Linguistic Compression in Single-Sentence Human-Written Summaries | findings-emnlp.532 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.533.bib | https://aclanthology.org/2023.findings-emnlp.533/ | @inproceedings{huang-etal-2023-mclf,
title = "{MCLF}: A Multi-grained Contrastive Learning Framework for {ASR}-robust Spoken Language Understanding",
author = "Huang, Zhiqi and
Chen, Dongsheng and
Zhu, Zhihong and
Cheng, Xuxin",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.533",
doi = "10.18653/v1/2023.findings-emnlp.533",
pages = "7936--7949",
abstract = "Enhancing the robustness towards Automatic Speech Recognition (ASR) errors is of great importance for Spoken Language Understanding (SLU). Trending ASR-robust SLU systems have witnessed impressive improvements through global contrastive learning. However, although most ASR errors occur only at local positions of utterances, they can easily lead to severe semantic changes, and utterance-level classification or comparison is difficult to distinguish such differences. To address the problem, we propose a two-stage multi-grained contrastive learning framework dubbed MCLF. Technically, we first adapt the pre-trained language models to downstream SLU datasets via the proposed multi-grained contrastive learning objective and then fine-tune it on the corresponding dataset. Besides, to facilitate contrastive learning in the pre-training stage, we explore several data augmentation methods to expand the training data. Experimental results and detailed analyses on four datasets and four BERT-like backbone models demonstrate the effectiveness of our approach.",
}
| Enhancing the robustness towards Automatic Speech Recognition (ASR) errors is of great importance for Spoken Language Understanding (SLU). Trending ASR-robust SLU systems have witnessed impressive improvements through global contrastive learning. However, although most ASR errors occur only at local positions of utterances, they can easily lead to severe semantic changes, and utterance-level classification or comparison is difficult to distinguish such differences. To address the problem, we propose a two-stage multi-grained contrastive learning framework dubbed MCLF. Technically, we first adapt the pre-trained language models to downstream SLU datasets via the proposed multi-grained contrastive learning objective and then fine-tune it on the corresponding dataset. Besides, to facilitate contrastive learning in the pre-training stage, we explore several data augmentation methods to expand the training data. Experimental results and detailed analyses on four datasets and four BERT-like backbone models demonstrate the effectiveness of our approach. | [
"Huang, Zhiqi",
"Chen, Dongsheng",
"Zhu, Zhihong",
"Cheng, Xuxin"
] | MCLF: A Multi-grained Contrastive Learning Framework for ASR-robust Spoken Language Understanding | findings-emnlp.533 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.534.bib | https://aclanthology.org/2023.findings-emnlp.534/ | @inproceedings{lim-etal-2023-beyond,
title = "Beyond Candidates : Adaptive Dialogue Agent Utilizing Persona and Knowledge",
author = "Lim, Jungwoo and
Kang, Myunghoon and
Kim, Jinsung and
Kim, Jeongwook and
Hur, Yuna and
Lim, Heuiseok",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.534",
doi = "10.18653/v1/2023.findings-emnlp.534",
pages = "7950--7963",
abstract = "To build ultimate dialogue agents, previous studies suggest models that ground both persona and knowledge. However, applying the dialogue system directly to the usual conversation is still limited because the system requires a complete sentence-formed persona and knowledge candidate sets from the given dataset. In contrast to the dialogue setting in the dataset, humans utilize semantic concepts in their minds rather than a set of pre-defined candidate sentences. Following this manner of human dialogue, we suggest an adaptive dialogue system that is applicable to situations where complete sentence-formed candidates are not given. Our model generates consistent and relevant persona descriptions and identifies relevant knowledge for engaging and knowledgeable responses, even with fragmentary information. We show that our model outperforms previous baselines that utilize persona and knowledge candidate sentences and conduct the human evaluation on the machine-generated responses. In addition, we conduct ablation studies to demonstrate the effectiveness of each component of our model. Furthermore, we apply our model to other dialogue datasets that only ground knowledge or persona to showcase its adaptability. Our code is available at https://github.com/dlawjddn803/BeCand.",
}
| To build ultimate dialogue agents, previous studies suggest models that ground both persona and knowledge. However, applying the dialogue system directly to the usual conversation is still limited because the system requires a complete sentence-formed persona and knowledge candidate sets from the given dataset. In contrast to the dialogue setting in the dataset, humans utilize semantic concepts in their minds rather than a set of pre-defined candidate sentences. Following this manner of human dialogue, we suggest an adaptive dialogue system that is applicable to situations where complete sentence-formed candidates are not given. Our model generates consistent and relevant persona descriptions and identifies relevant knowledge for engaging and knowledgeable responses, even with fragmentary information. We show that our model outperforms previous baselines that utilize persona and knowledge candidate sentences and conduct the human evaluation on the machine-generated responses. In addition, we conduct ablation studies to demonstrate the effectiveness of each component of our model. Furthermore, we apply our model to other dialogue datasets that only ground knowledge or persona to showcase its adaptability. Our code is available at https://github.com/dlawjddn803/BeCand. | [
"Lim, Jungwoo",
"Kang, Myunghoon",
"Kim, Jinsung",
"Kim, Jeongwook",
"Hur, Yuna",
"Lim, Heuiseok"
] | Beyond Candidates : Adaptive Dialogue Agent Utilizing Persona and Knowledge | findings-emnlp.534 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.535.bib | https://aclanthology.org/2023.findings-emnlp.535/ | @inproceedings{zhang-etal-2023-smartspanner,
title = "{S}mart{S}pan{NER}: Making {S}pan{NER} Robust in Low Resource Scenarios",
author = "Zhang, Min and
Qiao, Xiaosong and
Zhao, Yanqing and
Tao, Shimin and
Yang, Hao",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.535",
doi = "10.18653/v1/2023.findings-emnlp.535",
pages = "7964--7976",
abstract = "Named Entity Recognition (NER) is one of the most fundamental tasks in natural language processing. Span-level prediction (SpanNER) is more naturally suitable for nested NER than sequence labeling (SeqLab). However, according to our experiments, the SpanNER method is more sensitive to the amount of training data, i.e., the F1 score of SpanNER drops much more than that of SeqLab when the amount of training data drops. In order to improve the robustness of SpanNER in low resource scenarios, we propose a simple and effective method SmartSpanNER, which introduces a Named Entity Head (NEH) prediction task to SpanNER and performs multi-task learning together with the task of span classification. Experimental results demonstrate that the robustness of SpanNER could be greatly improved by SmartSpanNER in low resource scenarios constructed on the CoNLL03, Few-NERD, GENIA and ACE05 standard benchmark datasets.",
}
| Named Entity Recognition (NER) is one of the most fundamental tasks in natural language processing. Span-level prediction (SpanNER) is more naturally suitable for nested NER than sequence labeling (SeqLab). However, according to our experiments, the SpanNER method is more sensitive to the amount of training data, i.e., the F1 score of SpanNER drops much more than that of SeqLab when the amount of training data drops. In order to improve the robustness of SpanNER in low resource scenarios, we propose a simple and effective method SmartSpanNER, which introduces a Named Entity Head (NEH) prediction task to SpanNER and performs multi-task learning together with the task of span classification. Experimental results demonstrate that the robustness of SpanNER could be greatly improved by SmartSpanNER in low resource scenarios constructed on the CoNLL03, Few-NERD, GENIA and ACE05 standard benchmark datasets. | [
"Zhang, Min",
"Qiao, Xiaosong",
"Zhao, Yanqing",
"Tao, Shimin",
"Yang, Hao"
] | SmartSpanNER: Making SpanNER Robust in Low Resource Scenarios | findings-emnlp.535 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.536.bib | https://aclanthology.org/2023.findings-emnlp.536/ | @inproceedings{shaham-etal-2023-zeroscrolls,
title = "{Z}ero{SCROLLS}: A Zero-Shot Benchmark for Long Text Understanding",
author = "Shaham, Uri and
Ivgi, Maor and
Efrat, Avia and
Berant, Jonathan and
Levy, Omer",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.536",
doi = "10.18653/v1/2023.findings-emnlp.536",
pages = "7977--7989",
abstract = "We introduce ZeroSCROLLS, a zero-shot benchmark for natural language understanding over long texts, which contains only test and small validation sets, without training data. We adapt six tasks from the SCROLLS benchmark, and add four new datasets, including two novel information fusing tasks, such as aggregating the percentage of positive reviews. Using ZeroSCROLLS, we conduct a comprehensive evaluation of both open-source and closed large language models, finding that Claude outperforms ChatGPT, and that GPT-4 achieves the highest average score. However, there is still room for improvement on multiple open challenges in ZeroSCROLLS, such as aggregation tasks, where models struggle to pass the naive baseline. As the state of the art is a moving target, we invite researchers to evaluate their ideas on the live ZeroSCROLLS leaderboard.",
}
| We introduce ZeroSCROLLS, a zero-shot benchmark for natural language understanding over long texts, which contains only test and small validation sets, without training data. We adapt six tasks from the SCROLLS benchmark, and add four new datasets, including two novel information fusing tasks, such as aggregating the percentage of positive reviews. Using ZeroSCROLLS, we conduct a comprehensive evaluation of both open-source and closed large language models, finding that Claude outperforms ChatGPT, and that GPT-4 achieves the highest average score. However, there is still room for improvement on multiple open challenges in ZeroSCROLLS, such as aggregation tasks, where models struggle to pass the naive baseline. As the state of the art is a moving target, we invite researchers to evaluate their ideas on the live ZeroSCROLLS leaderboard. | [
"Shaham, Uri",
"Ivgi, Maor",
"Efrat, Avia",
"Berant, Jonathan",
"Levy, Omer"
] | ZeroSCROLLS: A Zero-Shot Benchmark for Long Text Understanding | findings-emnlp.536 | 2305.14196 | [
"https://github.com/tau-nlp/zero_scrolls"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.537.bib | https://aclanthology.org/2023.findings-emnlp.537/ | @inproceedings{sun-etal-2023-data,
title = "Data Selection Curriculum for Abstractive Text Summarization",
author = "Sun, Shichao and
Yuan, Ruifeng and
He, Jianfei and
Cao, Ziqiang and
Li, Wenjie and
Jia, Xiaohua",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.537",
doi = "10.18653/v1/2023.findings-emnlp.537",
pages = "7990--7995",
abstract = "Abstractive Text Summarization (ATS) models are commonly trained using large-scale data that is randomly shuffled. However, the impact of data selection and data ordering on ATS models remains a relatively unexplored research area, where a significant challenge lies in accurately assessing the learning difficulty of each training instance. This study introduces a Data Selection Curriculum (DSC) scoring system that incorporates both the difficulty of improving ATS model via an instance and the expected performance on this instance. By selectively excluding excessively simple and overly complex instances, the training efficiency can be optimized. Furthermore, curriculum learning is integrated to accelerate convergence and improve performance by gradually increasing the learning difficulty, inspired by human learners. Experimental results on the CNN/DailyMail dataset demonstrate that our approach surpasses potent baselines, utilizing a mere 20{\%} of the available instances.",
}
| Abstractive Text Summarization (ATS) models are commonly trained using large-scale data that is randomly shuffled. However, the impact of data selection and data ordering on ATS models remains a relatively unexplored research area, where a significant challenge lies in accurately assessing the learning difficulty of each training instance. This study introduces a Data Selection Curriculum (DSC) scoring system that incorporates both the difficulty of improving ATS model via an instance and the expected performance on this instance. By selectively excluding excessively simple and overly complex instances, the training efficiency can be optimized. Furthermore, curriculum learning is integrated to accelerate convergence and improve performance by gradually increasing the learning difficulty, inspired by human learners. Experimental results on the CNN/DailyMail dataset demonstrate that our approach surpasses potent baselines, utilizing a mere 20{\%} of the available instances. | [
"Sun, Shichao",
"Yuan, Ruifeng",
"He, Jianfei",
"Cao, Ziqiang",
"Li, Wenjie",
"Jia, Xiaohua"
] | Data Selection Curriculum for Abstractive Text Summarization | findings-emnlp.537 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.538.bib | https://aclanthology.org/2023.findings-emnlp.538/ | @inproceedings{purkayastha-etal-2023-romanization,
title = "{R}omanization-based Large-scale Adaptation of Multilingual Language Models",
author = "Purkayastha, Sukannya and
Ruder, Sebastian and
Pfeiffer, Jonas and
Gurevych, Iryna and
Vuli{\'c}, Ivan",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.538",
doi = "10.18653/v1/2023.findings-emnlp.538",
pages = "7996--8005",
abstract = "Large multilingual pretrained language models (mPLMs) have become the de facto state of the art for cross-lingual transfer in NLP. However, their large-scale deployment to many languages, besides pretraining data scarcity, is also hindered by the increase in vocabulary size and limitations in their parameter budget. In order to boost the capacity of mPLMs to deal with low-resource and unseen languages, we explore the potential of leveraging transliteration on a massive scale. In particular, we explore the UROMAN transliteration tool, which provides mappings from UTF-8 to Latin characters for all the writing systems, enabling inexpensive romanization for virtually any language. We first focus on establishing how UROMAN compares against other language-specific and manually curated transliterators for adapting multilingual PLMs. We then study and compare a plethora of data- and parameter-efficient strategies for adapting the mPLMs to romanized and non-romanized corpora of 14 diverse low-resource languages. Our results reveal that UROMAN-based transliteration can offer strong performance for many languages, with particular gains achieved in the most challenging setups: on languages with unseen scripts and with limited training data without any vocabulary augmentation. Further analyses reveal that an improved tokenizer based on romanized data can even outperform non-transliteration-based methods in the majority of languages.",
}
| Large multilingual pretrained language models (mPLMs) have become the de facto state of the art for cross-lingual transfer in NLP. However, their large-scale deployment to many languages, besides pretraining data scarcity, is also hindered by the increase in vocabulary size and limitations in their parameter budget. In order to boost the capacity of mPLMs to deal with low-resource and unseen languages, we explore the potential of leveraging transliteration on a massive scale. In particular, we explore the UROMAN transliteration tool, which provides mappings from UTF-8 to Latin characters for all the writing systems, enabling inexpensive romanization for virtually any language. We first focus on establishing how UROMAN compares against other language-specific and manually curated transliterators for adapting multilingual PLMs. We then study and compare a plethora of data- and parameter-efficient strategies for adapting the mPLMs to romanized and non-romanized corpora of 14 diverse low-resource languages. Our results reveal that UROMAN-based transliteration can offer strong performance for many languages, with particular gains achieved in the most challenging setups: on languages with unseen scripts and with limited training data without any vocabulary augmentation. Further analyses reveal that an improved tokenizer based on romanized data can even outperform non-transliteration-based methods in the majority of languages. | [
"Purkayastha, Sukannya",
"Ruder, Sebastian",
"Pfeiffer, Jonas",
"Gurevych, Iryna",
"Vuli{\\'c}, Ivan"
] | Romanization-based Large-scale Adaptation of Multilingual Language Models | findings-emnlp.538 | 2304.08865 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.539.bib | https://aclanthology.org/2023.findings-emnlp.539/ | @inproceedings{onorati-etal-2023-measuring,
title = "Measuring bias in Instruction-Following models with {P}-{AT}",
author = "Onorati, Dario and
Ruzzetti, Elena Sofia and
Venditti, Davide and
Ranaldi, Leonardo and
Zanzotto, Fabio Massimo",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.539",
doi = "10.18653/v1/2023.findings-emnlp.539",
pages = "8006--8034",
abstract = "Instruction-Following Language Models (IFLMs) are promising and versatile tools for solving many downstream, information-seeking tasks. Given their success, there is an urgent need to have a shared resource to determine whether existing and new IFLMs are prone to produce biased language interactions. In this paper, we propose Prompt Association Test (P-AT): a new resource for testing the presence of social biases in IFLMs. P-AT stems from WEAT (Caliskan et al., 2017) and generalizes the notion of measuring social biases to IFLMs. Basically, we cast WEAT word tests in promptized classification tasks, and we associate a metric - the bias score. Our resource consists of 2310 prompts. We then experimented with several families of IFLMs discovering gender and race biases in all the analyzed models. We expect P-AT to be an important tool for quantifying bias across different dimensions and, therefore, for encouraging the creation of fairer IFLMs before their distortions have consequences in the real world.",
}
| Instruction-Following Language Models (IFLMs) are promising and versatile tools for solving many downstream, information-seeking tasks. Given their success, there is an urgent need to have a shared resource to determine whether existing and new IFLMs are prone to produce biased language interactions. In this paper, we propose Prompt Association Test (P-AT): a new resource for testing the presence of social biases in IFLMs. P-AT stems from WEAT (Caliskan et al., 2017) and generalizes the notion of measuring social biases to IFLMs. Basically, we cast WEAT word tests in promptized classification tasks, and we associate a metric - the bias score. Our resource consists of 2310 prompts. We then experimented with several families of IFLMs discovering gender and race biases in all the analyzed models. We expect P-AT to be an important tool for quantifying bias across different dimensions and, therefore, for encouraging the creation of fairer IFLMs before their distortions have consequences in the real world. | [
"Onorati, Dario",
"Ruzzetti, Elena Sofia",
"Venditti, Davide",
"Ranaldi, Leonardo",
"Zanzotto, Fabio Massimo"
] | Measuring bias in Instruction-Following models with P-AT | findings-emnlp.539 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.540.bib | https://aclanthology.org/2023.findings-emnlp.540/ | @inproceedings{ling-etal-2023-open,
title = "Open-ended Commonsense Reasoning with Unrestricted Answer Candidates",
author = "Ling, Chen and
Zhang, Xuchao and
Zhao, Xujiang and
Liu, Yanchi and
Cheng, Wei and
Oishi, Mika and
Osaki, Takao and
Matsuda, Katsushi and
Chen, Haifeng and
Zhao, Liang",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.540",
doi = "10.18653/v1/2023.findings-emnlp.540",
pages = "8035--8047",
abstract = "Open-ended Commonsense Reasoning is defined as solving a commonsense question without providing 1) a short list of answer candidates and 2) a pre-defined answer scope. Conventional ways of formulating the commonsense question into a question-answering form or utilizing external knowledge to learn retrieval-based methods are less applicable in the open-ended setting due to an inherent challenge. Without pre-defining an answer scope or a few candidates, open-ended commonsense reasoning entails predicting answers by searching over an extremely large searching space. Moreover, most questions require implicit multi-hop reasoning, which presents even more challenges to our problem. In this work, we leverage pre-trained language models to iteratively retrieve reasoning paths on the external knowledge base, which does not require task-specific supervision. The reasoning paths can help to identify the most precise answer to the commonsense question. We conduct experiments on two commonsense benchmark datasets. Compared to other approaches, our proposed method achieves better performance both quantitatively and qualitatively.",
}
| Open-ended Commonsense Reasoning is defined as solving a commonsense question without providing 1) a short list of answer candidates and 2) a pre-defined answer scope. Conventional ways of formulating the commonsense question into a question-answering form or utilizing external knowledge to learn retrieval-based methods are less applicable in the open-ended setting due to an inherent challenge. Without pre-defining an answer scope or a few candidates, open-ended commonsense reasoning entails predicting answers by searching over an extremely large searching space. Moreover, most questions require implicit multi-hop reasoning, which presents even more challenges to our problem. In this work, we leverage pre-trained language models to iteratively retrieve reasoning paths on the external knowledge base, which does not require task-specific supervision. The reasoning paths can help to identify the most precise answer to the commonsense question. We conduct experiments on two commonsense benchmark datasets. Compared to other approaches, our proposed method achieves better performance both quantitatively and qualitatively. | [
"Ling, Chen",
"Zhang, Xuchao",
"Zhao, Xujiang",
"Liu, Yanchi",
"Cheng, Wei",
"Oishi, Mika",
"Osaki, Takao",
"Matsuda, Katsushi",
"Chen, Haifeng",
"Zhao, Liang"
] | Open-ended Commonsense Reasoning with Unrestricted Answer Candidates | findings-emnlp.540 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.541.bib | https://aclanthology.org/2023.findings-emnlp.541/ | @inproceedings{maimon-adi-2023-speaking,
title = "Speaking Style Conversion in the Waveform Domain Using Discrete Self-Supervised Units",
author = "Maimon, Gallil and
Adi, Yossi",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.541",
doi = "10.18653/v1/2023.findings-emnlp.541",
pages = "8048--8061",
abstract = "We introduce DISSC, a novel, lightweight method that converts the rhythm, pitch contour and timbre of a recording to a target speaker in a textless manner. Unlike DISSC, most voice conversion (VC) methods focus primarily on timbre, and ignore people{'}s unique speaking style (prosody). The proposed approach uses a pretrained, self-supervised model for encoding speech to discrete units, which makes it simple, effective, and fast to train. All conversion modules are only trained on reconstruction like tasks, thus suitable for any-to-many VC with no paired data. We introduce a suite of quantitative and qualitative evaluation metrics for this setup, and empirically demonstrate that DISSC significantly outperforms the evaluated baselines. Code and samples are available at https://pages.cs.huji.ac.il/adiyoss-lab/dissc/.",
}
| We introduce DISSC, a novel, lightweight method that converts the rhythm, pitch contour and timbre of a recording to a target speaker in a textless manner. Unlike DISSC, most voice conversion (VC) methods focus primarily on timbre, and ignore people{'}s unique speaking style (prosody). The proposed approach uses a pretrained, self-supervised model for encoding speech to discrete units, which makes it simple, effective, and fast to train. All conversion modules are only trained on reconstruction like tasks, thus suitable for any-to-many VC with no paired data. We introduce a suite of quantitative and qualitative evaluation metrics for this setup, and empirically demonstrate that DISSC significantly outperforms the evaluated baselines. Code and samples are available at https://pages.cs.huji.ac.il/adiyoss-lab/dissc/. | [
"Maimon, Gallil",
"Adi, Yossi"
] | Speaking Style Conversion in the Waveform Domain Using Discrete Self-Supervised Units | findings-emnlp.541 | 2212.09730 | [
"https://github.com/gallilmaimon/DISSC"
] | https://huggingface.co/papers/2212.09730 | 1 | 1 | 1 | 2 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.542.bib | https://aclanthology.org/2023.findings-emnlp.542/ | @inproceedings{liu-etal-2023-knowledge,
title = "Knowledge-Selective Pretraining for Attribute Value Extraction",
author = "Liu, Hui and
Yin, Qingyu and
Wang, Zhengyang and
Zhang, Chenwei and
Jiang, Haoming and
Gao, Yifan and
Li, Zheng and
Li, Xian and
Zhang, Chao and
Yin, Bing and
Wang, William and
Zhu, Xiaodan",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.542",
doi = "10.18653/v1/2023.findings-emnlp.542",
pages = "8062--8074",
abstract = "Attribute Value Extraction (AVE) aims to retrieve the values of attributes from the product profiles. The state-of-the-art methods tackle the AVE task through a question-answering (QA) paradigm, where the value is predicted from the context (i.e. product profile) given a query (i.e. attributes). Despite of the substantial advancements that have been made, the performance of existing methods on rare attributes is still far from satisfaction, and they cannot be easily extended to unseen attributes due to the poor generalization ability. In this work, we propose to leverage pretraining and transfer learning to address the aforementioned weaknesses. We first collect the product information from various E-commerce stores and retrieve a large number of (profile, attribute, value) triples, which will be used as the pretraining corpus. To more effectively utilize the retrieved corpus, we further design a Knowledge-Selective Framework (KSelF) based on query expansion that can be closely combined with the pretraining corpus to boost the performance. Meanwhile, considering the public AE-pub dataset contains considerable noise, we construct and contribute a larger benchmark EC-AVE collected from E-commerce websites. We conduct evaluation on both of these datasets. The experimental results demonstrate that our proposed KSelF achieves new state-of-the-art performance without pretraining. When incorporated with the pretraining corpus, the performance of KSelF can be further improved, particularly on the attributes with limited training resources.",
}
| Attribute Value Extraction (AVE) aims to retrieve the values of attributes from the product profiles. The state-of-the-art methods tackle the AVE task through a question-answering (QA) paradigm, where the value is predicted from the context (i.e. product profile) given a query (i.e. attributes). Despite of the substantial advancements that have been made, the performance of existing methods on rare attributes is still far from satisfaction, and they cannot be easily extended to unseen attributes due to the poor generalization ability. In this work, we propose to leverage pretraining and transfer learning to address the aforementioned weaknesses. We first collect the product information from various E-commerce stores and retrieve a large number of (profile, attribute, value) triples, which will be used as the pretraining corpus. To more effectively utilize the retrieved corpus, we further design a Knowledge-Selective Framework (KSelF) based on query expansion that can be closely combined with the pretraining corpus to boost the performance. Meanwhile, considering the public AE-pub dataset contains considerable noise, we construct and contribute a larger benchmark EC-AVE collected from E-commerce websites. We conduct evaluation on both of these datasets. The experimental results demonstrate that our proposed KSelF achieves new state-of-the-art performance without pretraining. When incorporated with the pretraining corpus, the performance of KSelF can be further improved, particularly on the attributes with limited training resources. | [
"Liu, Hui",
"Yin, Qingyu",
"Wang, Zhengyang",
"Zhang, Chenwei",
"Jiang, Haoming",
"Gao, Yifan",
"Li, Zheng",
"Li, Xian",
"Zhang, Chao",
"Yin, Bing",
"Wang, William",
"Zhu, Xiaodan"
] | Knowledge-Selective Pretraining for Attribute Value Extraction | findings-emnlp.542 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.543.bib | https://aclanthology.org/2023.findings-emnlp.543/ | @inproceedings{wan-etal-2023-new,
title = "New Datasets and Controllable Iterative Data Augmentation Method for Code-switching {ASR} Error Correction",
author = "Wan, Zhaohong and
Wan, Xiaojun and
Peng, Wei and
Li, Rongjun",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.543",
doi = "10.18653/v1/2023.findings-emnlp.543",
pages = "8075--8087",
abstract = "With the wide use of automatic speech recognition(ASR) systems, researchers pay more attention to the ASR error correction task to improve the quality of recognition results. In particular, ASR in bilingual or multilingual settings, namely code-switching ASR, has greater challenges and research value. In this paper, we first present code-switching ASR correction datasets obtained from solid ASR systems and automatic annotators. The datasets contain Chinese-English code-switching dialogues of bilingual speakers in Singapore, Malaysia, and Hong Kong. Based on this task, we propose a controllable iterative (CI) data augmentation method for improving the performance of mainstream ASR error correction systems. With a small amount of training data, our proposed method has the ability to iteratively produce abundant pseudo parallel data from the monolingual corpus for Chinese-English code-switching ASR correction. Results of experiments show that our method achieves the best performance compared with the rule-based, back-translation-based data augmentation methods and large language model ChatGPT.",
}
| With the wide use of automatic speech recognition(ASR) systems, researchers pay more attention to the ASR error correction task to improve the quality of recognition results. In particular, ASR in bilingual or multilingual settings, namely code-switching ASR, has greater challenges and research value. In this paper, we first present code-switching ASR correction datasets obtained from solid ASR systems and automatic annotators. The datasets contain Chinese-English code-switching dialogues of bilingual speakers in Singapore, Malaysia, and Hong Kong. Based on this task, we propose a controllable iterative (CI) data augmentation method for improving the performance of mainstream ASR error correction systems. With a small amount of training data, our proposed method has the ability to iteratively produce abundant pseudo parallel data from the monolingual corpus for Chinese-English code-switching ASR correction. Results of experiments show that our method achieves the best performance compared with the rule-based, back-translation-based data augmentation methods and large language model ChatGPT. | [
"Wan, Zhaohong",
"Wan, Xiaojun",
"Peng, Wei",
"Li, Rongjun"
] | New Datasets and Controllable Iterative Data Augmentation Method for Code-switching ASR Error Correction | findings-emnlp.543 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.544.bib | https://aclanthology.org/2023.findings-emnlp.544/ | @inproceedings{yadav-etal-2023-efficient,
title = "Efficient k-{NN} Search with Cross-Encoders using Adaptive Multi-Round {CUR} Decomposition",
author = "Yadav, Nishant and
Monath, Nicholas and
Zaheer, Manzil and
McCallum, Andrew",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.544",
doi = "10.18653/v1/2023.findings-emnlp.544",
pages = "8088--8103",
abstract = "Cross-encoder models, which jointly encode and score a query-item pair, are prohibitively expensive for direct k-nearest neighbor (k-NN) search. Consequently, k-NN search typically employs a fast approximate retrieval (e.g. using BM25 or dual-encoder vectors), followed by reranking with a cross-encoder; however, the retrieval approximation often has detrimental recall regret. This problem is tackled by ANNCUR (Yadav et al., 2022), a recent work that employs a cross-encoder only, making search efficient using a relatively small number of anchor items, and a CUR matrix factorization. While ANNCUR{'}s one-time selection of anchors tends to approximate the cross-encoder distances on average, doing so forfeits the capacity to accurately estimate distances to items near the query, leading to regret in the crucial end-task: recall of top-k items. In this paper, we propose ADACUR, a method that adaptively, iteratively, and efficiently minimizes the approximation error for the practically important top-k neighbors. It does so by iteratively performing k-NN search using the anchors available so far, then adding these retrieved nearest neighbors to the anchor set for the next round. Empirically, on multiple datasets, in comparison to previous traditional and state-of-the-art methods such as ANNCUR and dual-encoder-based retrieve-and-rerank, our proposed approach ADACUR consistently reduces recall error{---}by up to 70{\%} on the important k = 1 setting{---}while using no more compute than its competitors.",
}
| Cross-encoder models, which jointly encode and score a query-item pair, are prohibitively expensive for direct k-nearest neighbor (k-NN) search. Consequently, k-NN search typically employs a fast approximate retrieval (e.g. using BM25 or dual-encoder vectors), followed by reranking with a cross-encoder; however, the retrieval approximation often has detrimental recall regret. This problem is tackled by ANNCUR (Yadav et al., 2022), a recent work that employs a cross-encoder only, making search efficient using a relatively small number of anchor items, and a CUR matrix factorization. While ANNCUR{'}s one-time selection of anchors tends to approximate the cross-encoder distances on average, doing so forfeits the capacity to accurately estimate distances to items near the query, leading to regret in the crucial end-task: recall of top-k items. In this paper, we propose ADACUR, a method that adaptively, iteratively, and efficiently minimizes the approximation error for the practically important top-k neighbors. It does so by iteratively performing k-NN search using the anchors available so far, then adding these retrieved nearest neighbors to the anchor set for the next round. Empirically, on multiple datasets, in comparison to previous traditional and state-of-the-art methods such as ANNCUR and dual-encoder-based retrieve-and-rerank, our proposed approach ADACUR consistently reduces recall error{---}by up to 70{\%} on the important k = 1 setting{---}while using no more compute than its competitors. | [
"Yadav, Nishant",
"Monath, Nicholas",
"Zaheer, Manzil",
"McCallum, Andrew"
] | Efficient k-NN Search with Cross-Encoders using Adaptive Multi-Round CUR Decomposition | findings-emnlp.544 | 2305.02996 | [
"https://github.com/iesl/anncur"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.545.bib | https://aclanthology.org/2023.findings-emnlp.545/ | @inproceedings{ji-etal-2023-isotropic,
title = "Isotropic Representation Can Improve Zero-Shot Cross-Lingual Transfer on Multilingual Language Models",
author = "Ji, Yixin and
Wang, Jikai and
Li, Juntao and
Ye, Hai and
Zhang, Min",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.545",
doi = "10.18653/v1/2023.findings-emnlp.545",
pages = "8104--8118",
abstract = "With the development of multilingual pre-trained language models (mPLMs), zero-shot cross-lingual transfer shows great potential. To further improve the performance of cross-lingual transfer, many studies have explored representation misalignment caused by morphological differences but neglected the misalignment caused by the anisotropic distribution of contextual representations. In this work, we propose enhanced isotropy and constrained code-switching for zero-shot cross-lingual transfer to alleviate the problem of misalignment caused by the anisotropic representations and maintain syntactic structural knowledge. Extensive experiments on three zero-shot cross-lingual transfer tasks demonstrate that our method gains significant improvements over strong mPLM backbones and further improves the state-of-the-art methods.",
}
| With the development of multilingual pre-trained language models (mPLMs), zero-shot cross-lingual transfer shows great potential. To further improve the performance of cross-lingual transfer, many studies have explored representation misalignment caused by morphological differences but neglected the misalignment caused by the anisotropic distribution of contextual representations. In this work, we propose enhanced isotropy and constrained code-switching for zero-shot cross-lingual transfer to alleviate the problem of misalignment caused by the anisotropic representations and maintain syntactic structural knowledge. Extensive experiments on three zero-shot cross-lingual transfer tasks demonstrate that our method gains significant improvements over strong mPLM backbones and further improves the state-of-the-art methods. | [
"Ji, Yixin",
"Wang, Jikai",
"Li, Juntao",
"Ye, Hai",
"Zhang, Min"
] | Isotropic Representation Can Improve Zero-Shot Cross-Lingual Transfer on Multilingual Language Models | findings-emnlp.545 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.546.bib | https://aclanthology.org/2023.findings-emnlp.546/ | @inproceedings{merlo-2023-blackbird,
title = "Blackbird language matrices ({BLM}), a new task for rule-like generalization in neural networks: Can Large Language Models pass the test?",
author = "Merlo, Paola",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.546",
doi = "10.18653/v1/2023.findings-emnlp.546",
pages = "8119--8152",
abstract = "How do we evaluate Large Language Models (LLMs) and determine the aspects and limits of their intelligent behaviour? It is currently conjectured that shortcomings of LLMs in multi-linguality and reasoning are due to a lack of ability to generalize. It has been argued that, instead, humans are better at generalization because they have a tendency at extracting rules from complex data. We propose a method to evaluate LLMs ability to rule-based generalization. When exposed to tests of analytic intelligence, for example the visual RAVEN IQ test, human problem-solvers identify the relevant objects in the picture and their relevant attributes and reason based on rules applied to them. Based on the induced rules, they are able to provide a generalisation and a solution to the test. An analogous language task has recently been proposed (called BLM) for LLM. In this paper, we argue that we can use this task to investigate what linguistic reasoning LLM develop, by asking them to solve some simple variants of the BLM task. We find that current state-of-the-art generative models, such as ChatGPT, can handle the task in the sense that they easily understand the instructions and can provide step-by-step reasoning that shows that it can solve two of the main cognitive hurdles: correspondence finding (object and attribute identification) and item novelty. However, overall they cannot find the correct answer, even with considerable help. In particular, they never identify the structure of the problem, exhibiting, we hypothesize, a lack of goal and subgoal management abilities, an ability that has been argued to measure differential abilities in humans. We argue that this finding supports the usefulness of the task as a method to test the limits and specific properties of generalisation ability in Large Language Models, providing an intrinsic evaluation method inspired by tests of human intelligence.",
}
| How do we evaluate Large Language Models (LLMs) and determine the aspects and limits of their intelligent behaviour? It is currently conjectured that shortcomings of LLMs in multi-linguality and reasoning are due to a lack of ability to generalize. It has been argued that, instead, humans are better at generalization because they have a tendency at extracting rules from complex data. We propose a method to evaluate LLMs ability to rule-based generalization. When exposed to tests of analytic intelligence, for example the visual RAVEN IQ test, human problem-solvers identify the relevant objects in the picture and their relevant attributes and reason based on rules applied to them. Based on the induced rules, they are able to provide a generalisation and a solution to the test. An analogous language task has recently been proposed (called BLM) for LLM. In this paper, we argue that we can use this task to investigate what linguistic reasoning LLM develop, by asking them to solve some simple variants of the BLM task. We find that current state-of-the-art generative models, such as ChatGPT, can handle the task in the sense that they easily understand the instructions and can provide step-by-step reasoning that shows that it can solve two of the main cognitive hurdles: correspondence finding (object and attribute identification) and item novelty. However, overall they cannot find the correct answer, even with considerable help. In particular, they never identify the structure of the problem, exhibiting, we hypothesize, a lack of goal and subgoal management abilities, an ability that has been argued to measure differential abilities in humans. We argue that this finding supports the usefulness of the task as a method to test the limits and specific properties of generalisation ability in Large Language Models, providing an intrinsic evaluation method inspired by tests of human intelligence. | [
"Merlo, Paola"
] | Blackbird language matrices (BLM), a new task for rule-like generalization in neural networks: Can Large Language Models pass the test? | findings-emnlp.546 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.547.bib | https://aclanthology.org/2023.findings-emnlp.547/ | @inproceedings{xu-etal-2023-distillcse,
title = "{D}istill{CSE}: Distilled Contrastive Learning for Sentence Embeddings",
author = "Xu, Jiahao and
Shao, Wei and
Chen, Lihui and
Liu, Lemao",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.547",
doi = "10.18653/v1/2023.findings-emnlp.547",
pages = "8153--8165",
abstract = "This paper proposes the DistillCSE framework, which performs contrastive learning under the self-training paradigm with knowledge distillation. The potential advantage of DistillCSE is its self-enhancing feature: using a base model to provide additional supervision signals, a stronger model may be learned through knowledge distillation. However, the vanilla DistillCSE through the standard implementation of knowledge distillation only achieves marginal improvements. The quantitative analyses demonstrate its reason that the standard knowledge distillation exhibits a relatively large variance of the teacher model{'}s logits due to the essence of contrastive learning. To mitigate the issue induced by high variance, this paper accordingly proposed two simple yet effective solutions for knowledge distillation: a Group-P shuffling strategy as an implicit regularization and the averaging logits from multiple teacher components. Experiments on standard benchmarks demonstrate that the proposed DistillCSE outperforms many strong baseline methods and yields a new state-of-the-art performance.",
}
| This paper proposes the DistillCSE framework, which performs contrastive learning under the self-training paradigm with knowledge distillation. The potential advantage of DistillCSE is its self-enhancing feature: using a base model to provide additional supervision signals, a stronger model may be learned through knowledge distillation. However, the vanilla DistillCSE through the standard implementation of knowledge distillation only achieves marginal improvements. The quantitative analyses demonstrate its reason that the standard knowledge distillation exhibits a relatively large variance of the teacher model{'}s logits due to the essence of contrastive learning. To mitigate the issue induced by high variance, this paper accordingly proposed two simple yet effective solutions for knowledge distillation: a Group-P shuffling strategy as an implicit regularization and the averaging logits from multiple teacher components. Experiments on standard benchmarks demonstrate that the proposed DistillCSE outperforms many strong baseline methods and yields a new state-of-the-art performance. | [
"Xu, Jiahao",
"Shao, Wei",
"Chen, Lihui",
"Liu, Lemao"
] | DistillCSE: Distilled Contrastive Learning for Sentence Embeddings | findings-emnlp.547 | 2310.13499 | [
"https://github.com/jiahao004/distillcse"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.548.bib | https://aclanthology.org/2023.findings-emnlp.548/ | @inproceedings{otto-etal-2023-gsap,
title = "{GSAP}-{NER}: A Novel Task, Corpus, and Baseline for Scholarly Entity Extraction Focused on Machine Learning Models and Datasets",
author = {Otto, Wolfgang and
Zloch, Matth{\"a}us and
Gan, Lu and
Karmakar, Saurav and
Dietze, Stefan},
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.548",
doi = "10.18653/v1/2023.findings-emnlp.548",
pages = "8166--8176",
abstract = "Named Entity Recognition (NER) models play a crucial role in various NLP tasks, including information extraction (IE) and text understanding. In academic writing, references to machine learning models and datasets are fundamental components of various computer science publications and necessitate accurate models for identification. Despite the advancements in NER, existing ground truth datasets do not treat fine-grained types like ML model and model architecture as separate entity types, and consequently, baseline models cannot recognize them as such. In this paper, we release a corpus of 100 manually annotated full-text scientific publications and a first baseline model for 10 entity types centered around ML models and datasets. In order to provide a nuanced understanding of how ML models and datasets are mentioned and utilized, our dataset also contains annotations for informal mentions like {``}our BERT-based model{''} or {``}an image CNN{''}. You can find the ground truth dataset and code to replicate model training at https://data.gesis.org/gsap/gsap-ner.",
}
| Named Entity Recognition (NER) models play a crucial role in various NLP tasks, including information extraction (IE) and text understanding. In academic writing, references to machine learning models and datasets are fundamental components of various computer science publications and necessitate accurate models for identification. Despite the advancements in NER, existing ground truth datasets do not treat fine-grained types like ML model and model architecture as separate entity types, and consequently, baseline models cannot recognize them as such. In this paper, we release a corpus of 100 manually annotated full-text scientific publications and a first baseline model for 10 entity types centered around ML models and datasets. In order to provide a nuanced understanding of how ML models and datasets are mentioned and utilized, our dataset also contains annotations for informal mentions like {``}our BERT-based model{''} or {``}an image CNN{''}. You can find the ground truth dataset and code to replicate model training at https://data.gesis.org/gsap/gsap-ner. | [
"Otto, Wolfgang",
"Zloch, Matth{\\\"a}us",
"Gan, Lu",
"Karmakar, Saurav",
"Dietze, Stefan"
] | GSAP-NER: A Novel Task, Corpus, and Baseline for Scholarly Entity Extraction Focused on Machine Learning Models and Datasets | findings-emnlp.548 | 2311.09860 | [
""
] | https://huggingface.co/papers/2311.09860 | 1 | 5 | 3 | 5 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.549.bib | https://aclanthology.org/2023.findings-emnlp.549/ | @inproceedings{giorgi-etal-2023-open,
title = "Open Domain Multi-document Summarization: A Comprehensive Study of Model Brittleness under Retrieval",
author = "Giorgi, John and
Soldaini, Luca and
Wang, Bo and
Bader, Gary and
Lo, Kyle and
Wang, Lucy and
Cohan, Arman",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.549",
doi = "10.18653/v1/2023.findings-emnlp.549",
pages = "8177--8199",
abstract = "Multi-document summarization (MDS) assumes a set of topic-related documents are provided as input. In practice, this document set is not always available; it would need to be retrieved given an information need, i.e. a question or topic statement, a setting we dub {``}open-domain{'} MDS. We study this more challenging setting by formalizing the task and bootstrapping it using existing datasets, retrievers and summarizers. Via extensive automatic and human evaluation, we determine: (1) state-of-the-art summarizers suffer large reductions in performance when applied to open-domain MDS, (2) additional training in the open-domain setting can reduce this sensitivity to imperfect retrieval, and (3) summarizers are insensitive to the retrieval of duplicate documents and the order of retrieved documents, but highly sensitive to other errors, like the retrieval of irrelevant documents. Based on our results, we provide practical guidelines to enable future work on open-domain MDS, e.g. how to choose the number of retrieved documents to summarize. Our results suggest that new retrieval and summarization methods and annotated resources for training and evaluation are necessary for further progress in the open-domain setting.",
}
| Multi-document summarization (MDS) assumes a set of topic-related documents are provided as input. In practice, this document set is not always available; it would need to be retrieved given an information need, i.e. a question or topic statement, a setting we dub {``}open-domain{'} MDS. We study this more challenging setting by formalizing the task and bootstrapping it using existing datasets, retrievers and summarizers. Via extensive automatic and human evaluation, we determine: (1) state-of-the-art summarizers suffer large reductions in performance when applied to open-domain MDS, (2) additional training in the open-domain setting can reduce this sensitivity to imperfect retrieval, and (3) summarizers are insensitive to the retrieval of duplicate documents and the order of retrieved documents, but highly sensitive to other errors, like the retrieval of irrelevant documents. Based on our results, we provide practical guidelines to enable future work on open-domain MDS, e.g. how to choose the number of retrieved documents to summarize. Our results suggest that new retrieval and summarization methods and annotated resources for training and evaluation are necessary for further progress in the open-domain setting. | [
"Giorgi, John",
"Soldaini, Luca",
"Wang, Bo",
"Bader, Gary",
"Lo, Kyle",
"Wang, Lucy",
"Cohan, Arman"
] | Open Domain Multi-document Summarization: A Comprehensive Study of Model Brittleness under Retrieval | findings-emnlp.549 | 2212.10526 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.550.bib | https://aclanthology.org/2023.findings-emnlp.550/ | @inproceedings{bansal-etal-2023-shot,
title = "Few-shot Unified Question Answering: Tuning Models or Prompts?",
author = "Bansal, Srijan and
Yavuz, Semih and
Pang, Bo and
Bhat, Meghana and
Zhou, Yingbo",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.550",
doi = "10.18653/v1/2023.findings-emnlp.550",
pages = "8200--8220",
abstract = "Question-answering (QA) tasks often investigate specific question types, knowledge domains, or reasoning skills, leading to specialized models catering to specific categories of QA tasks. While recent research has explored the idea of unified QA models, such models are usually explored for high-resource scenarios and require re-training to extend their capabilities. To overcome these drawbacks, the paper explores the potential of two paradigms of tuning, model, and prompts, for unified QA under a low-resource setting. The paper provides an exhaustive analysis of their applicability using 16 QA datasets, revealing that prompt tuning can perform as well as model tuning in a few-shot setting with a good initialization. The study also shows that parameter-sharing results in superior few-shot performance, simple knowledge transfer techniques for prompt initialization can be effective, and prompt tuning achieves a significant performance boost from pre-training in a low-resource regime. The research offers insights into the advantages and limitations of prompt tuning for unified QA in a few-shot setting, contributing to the development of effective and efficient systems in low-resource scenarios.",
}
| Question-answering (QA) tasks often investigate specific question types, knowledge domains, or reasoning skills, leading to specialized models catering to specific categories of QA tasks. While recent research has explored the idea of unified QA models, such models are usually explored for high-resource scenarios and require re-training to extend their capabilities. To overcome these drawbacks, the paper explores the potential of two paradigms of tuning, model, and prompts, for unified QA under a low-resource setting. The paper provides an exhaustive analysis of their applicability using 16 QA datasets, revealing that prompt tuning can perform as well as model tuning in a few-shot setting with a good initialization. The study also shows that parameter-sharing results in superior few-shot performance, simple knowledge transfer techniques for prompt initialization can be effective, and prompt tuning achieves a significant performance boost from pre-training in a low-resource regime. The research offers insights into the advantages and limitations of prompt tuning for unified QA in a few-shot setting, contributing to the development of effective and efficient systems in low-resource scenarios. | [
"Bansal, Srijan",
"Yavuz, Semih",
"Pang, Bo",
"Bhat, Meghana",
"Zhou, Yingbo"
] | Few-shot Unified Question Answering: Tuning Models or Prompts? | findings-emnlp.550 | 2305.14569 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.551.bib | https://aclanthology.org/2023.findings-emnlp.551/ | @inproceedings{markowska-etal-2023-finding,
title = "Finding Common Ground: Annotating and Predicting Common Ground in Spoken Conversations",
author = "Markowska, Magdalena and
Taghizadeh, Mohammad and
Soubki, Adil and
Mirroshandel, Seyed and
Rambow, Owen",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.551",
doi = "10.18653/v1/2023.findings-emnlp.551",
pages = "8221--8233",
abstract = "When we communicate with other humans, we do not simply generate a sequence of words. Rather, we use our cognitive state (beliefs, desires, intentions) and our model of the audience{'}s cognitive state to create utterances that affect the audience{'}s cognitive state in the intended manner. An important part of cognitive state is the common ground, which is the content the speaker believes, and the speaker believes the audience believes, and so on. While much attention has been paid to common ground in cognitive science, there has not been much work in natural language processing. In this paper, we introduce a new annotation and corpus to capture common ground. We then describe some initial experiments extracting propositions from dialog and tracking their status in the common ground from the perspective of each speaker.",
}
| When we communicate with other humans, we do not simply generate a sequence of words. Rather, we use our cognitive state (beliefs, desires, intentions) and our model of the audience{'}s cognitive state to create utterances that affect the audience{'}s cognitive state in the intended manner. An important part of cognitive state is the common ground, which is the content the speaker believes, and the speaker believes the audience believes, and so on. While much attention has been paid to common ground in cognitive science, there has not been much work in natural language processing. In this paper, we introduce a new annotation and corpus to capture common ground. We then describe some initial experiments extracting propositions from dialog and tracking their status in the common ground from the perspective of each speaker. | [
"Markowska, Magdalena",
"Taghizadeh, Mohammad",
"Soubki, Adil",
"Mirrosh",
"el, Seyed",
"Rambow, Owen"
] | Finding Common Ground: Annotating and Predicting Common Ground in Spoken Conversations | findings-emnlp.551 | 2311.01273 | [
"https://github.com/cogstates/2023-emnlp-common-ground"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.552.bib | https://aclanthology.org/2023.findings-emnlp.552/ | @inproceedings{si-etal-2023-getting,
title = "Getting {M}o{RE} out of Mixture of Language Model Reasoning Experts",
author = "Si, Chenglei and
Shi, Weijia and
Zhao, Chen and
Zettlemoyer, Luke and
Boyd-Graber, Jordan",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.552",
doi = "10.18653/v1/2023.findings-emnlp.552",
pages = "8234--8249",
abstract = "While recent large language models (LLMs) improve on various question answering (QA) datasets, it remains difficult for a single model to generalize across question types that require distinct reasoning abilities. We provide empirical evidence that state-of-the-art LLMs suffer from poor generalizability on reasoning types beyond those seen in the prompt. To remedy this, we propose a Mixture-of-Reasoning-Experts (MORE) framework that ensembles diverse specialized language models. We specialize the backbone language model with prompts optimized for different reasoning categories, including factual, multihop, mathematical, and commonsense reasoning. Our key insight is to leverage agreement among the specialized experts to select the best answer for each question, or to abstain from answering. This gives MORE higher accuracy than any single specialized model on a collection of 12 QA datasets from four reasoning types. Beyond generalizability, the interpretable design of MORE improves selective question answering results compared to baselines without incorporating inter-expert agreement. This framework is also more interpretable and useful to human consumers of QA outputs. Our human study confirms that presenting expert predictions and the answer selection process helps annotators more accurately calibrate when to trust the system{'}s output. We release all code and data to facilitate future work.",
}
| While recent large language models (LLMs) improve on various question answering (QA) datasets, it remains difficult for a single model to generalize across question types that require distinct reasoning abilities. We provide empirical evidence that state-of-the-art LLMs suffer from poor generalizability on reasoning types beyond those seen in the prompt. To remedy this, we propose a Mixture-of-Reasoning-Experts (MORE) framework that ensembles diverse specialized language models. We specialize the backbone language model with prompts optimized for different reasoning categories, including factual, multihop, mathematical, and commonsense reasoning. Our key insight is to leverage agreement among the specialized experts to select the best answer for each question, or to abstain from answering. This gives MORE higher accuracy than any single specialized model on a collection of 12 QA datasets from four reasoning types. Beyond generalizability, the interpretable design of MORE improves selective question answering results compared to baselines without incorporating inter-expert agreement. This framework is also more interpretable and useful to human consumers of QA outputs. Our human study confirms that presenting expert predictions and the answer selection process helps annotators more accurately calibrate when to trust the system{'}s output. We release all code and data to facilitate future work. | [
"Si, Chenglei",
"Shi, Weijia",
"Zhao, Chen",
"Zettlemoyer, Luke",
"Boyd-Graber, Jordan"
] | Getting MoRE out of Mixture of Language Model Reasoning Experts | findings-emnlp.552 | 2305.14628 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.553.bib | https://aclanthology.org/2023.findings-emnlp.553/ | @inproceedings{ettinger-etal-2023-expert,
title = "{``}You Are An Expert Linguistic Annotator{''}: Limits of {LLM}s as Analyzers of {A}bstract {M}eaning {R}epresentation",
author = "Ettinger, Allyson and
Hwang, Jena and
Pyatkin, Valentina and
Bhagavatula, Chandra and
Choi, Yejin",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.553",
doi = "10.18653/v1/2023.findings-emnlp.553",
pages = "8250--8263",
abstract = "Large language models (LLMs) demonstrate an amazing proficiency and fluency in the $\textit{use}$ of language. Does that mean that they have also acquired insightful linguistic knowledge $\textit{about}$ the language, to an extent that they can serve as an {``}expert linguistic annotator{'}? In this paper, we examine the successes and limitations of the GPT-3, ChatGPT, and GPT-4 models, focusing on the Abstract Meaning Representation (AMR) parsing formalism (Banarescu et al., 2013), which provides rich graphical representations of sentence meaning structure while abstracting away from surface forms. We compare models{'} analysis of this semantic structure across two settings: 1) direct production of AMR parses based on zero- and few-shot examples, and 2) indirect partial reconstruction of AMR via metalinguistic natural language queries (e.g., {``}Identify the primary event of this sentence, and the predicate corresponding to that event.{''}). Across these settings, we find that models can reliably reproduce the basic format of AMR, as well as some core event, argument, and modifier structure$-$however, model outputs are prone to frequent and major errors, and holistic analysis of parse acceptability shows that even with few-shot demonstrations, models have virtually 0{\%} success in producing fully accurate parses. Eliciting responses in natural language produces similar patterns of errors. Overall, our findings indicate that these models out-of-the-box can accurately identify some core aspects of semantic structure, but there remain key limitations in their ability to support fully accurate semantic analyses or parses.",
}
| Large language models (LLMs) demonstrate an amazing proficiency and fluency in the $\textit{use}$ of language. Does that mean that they have also acquired insightful linguistic knowledge $\textit{about}$ the language, to an extent that they can serve as an {``}expert linguistic annotator{'}? In this paper, we examine the successes and limitations of the GPT-3, ChatGPT, and GPT-4 models, focusing on the Abstract Meaning Representation (AMR) parsing formalism (Banarescu et al., 2013), which provides rich graphical representations of sentence meaning structure while abstracting away from surface forms. We compare models{'} analysis of this semantic structure across two settings: 1) direct production of AMR parses based on zero- and few-shot examples, and 2) indirect partial reconstruction of AMR via metalinguistic natural language queries (e.g., {``}Identify the primary event of this sentence, and the predicate corresponding to that event.{''}). Across these settings, we find that models can reliably reproduce the basic format of AMR, as well as some core event, argument, and modifier structure$-$however, model outputs are prone to frequent and major errors, and holistic analysis of parse acceptability shows that even with few-shot demonstrations, models have virtually 0{\%} success in producing fully accurate parses. Eliciting responses in natural language produces similar patterns of errors. Overall, our findings indicate that these models out-of-the-box can accurately identify some core aspects of semantic structure, but there remain key limitations in their ability to support fully accurate semantic analyses or parses. | [
"Ettinger, Allyson",
"Hwang, Jena",
"Pyatkin, Valentina",
"Bhagavatula, Ch",
"ra",
"Choi, Yejin"
] | “You Are An Expert Linguistic Annotator”: Limits of LLMs as Analyzers of Abstract Meaning Representation | findings-emnlp.553 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.554.bib | https://aclanthology.org/2023.findings-emnlp.554/ | @inproceedings{basile-etal-2023-zero,
title = "Zero-Shot Data Maps. Efficient Dataset Cartography Without Model Training",
author = "Basile, Angelo and
Franco-Salvador, Marc and
Rosso, Paolo",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.554",
doi = "10.18653/v1/2023.findings-emnlp.554",
pages = "8264--8277",
abstract = "Data Maps (Swayamdipta, et al. 2020) have emerged as a powerful tool for diagnosing large annotated datasets. Given a model fitted on a dataset, these maps show each data instance from the dataset in a 2-dimensional space defined by a) the model{'}s confidence in the true class and b) the variability of this confidence. In previous work, confidence and variability are usually computed using training dynamics, which requires the fitting of a strong model to the dataset. In this paper, we introduce a novel approach: Zero-Shot Data Maps based on fast bi-encoder networks. For each data point, confidence on the true label and variability are computed over the members of an ensemble of zero-shot models constructed with different {---} but semantically equivalent {---} label descriptions, i.e., textual representations of each class in a given label space. We conduct a comparative analysis of maps compiled using traditional training dynamics and our proposed zero-shot models across various datasets. Our findings reveal that Zero-Shot Data Maps generally match those produced by the traditional method while delivering up to a 14x speedup. The code is available [here](https://github.com/symanto-research/zeroshot-cartography).",
}
| Data Maps (Swayamdipta, et al. 2020) have emerged as a powerful tool for diagnosing large annotated datasets. Given a model fitted on a dataset, these maps show each data instance from the dataset in a 2-dimensional space defined by a) the model{'}s confidence in the true class and b) the variability of this confidence. In previous work, confidence and variability are usually computed using training dynamics, which requires the fitting of a strong model to the dataset. In this paper, we introduce a novel approach: Zero-Shot Data Maps based on fast bi-encoder networks. For each data point, confidence on the true label and variability are computed over the members of an ensemble of zero-shot models constructed with different {---} but semantically equivalent {---} label descriptions, i.e., textual representations of each class in a given label space. We conduct a comparative analysis of maps compiled using traditional training dynamics and our proposed zero-shot models across various datasets. Our findings reveal that Zero-Shot Data Maps generally match those produced by the traditional method while delivering up to a 14x speedup. The code is available [here](https://github.com/symanto-research/zeroshot-cartography). | [
"Basile, Angelo",
"Franco-Salvador, Marc",
"Rosso, Paolo"
] | Zero-Shot Data Maps. Efficient Dataset Cartography Without Model Training | findings-emnlp.554 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.555.bib | https://aclanthology.org/2023.findings-emnlp.555/ | @inproceedings{guo-etal-2023-isotropy,
title = "Isotropy-Enhanced Conditional Masked Language Models",
author = "Guo, Pei and
Xiao, Yisheng and
Li, Juntao and
Ji, Yixin and
Zhang, Min",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.555",
doi = "10.18653/v1/2023.findings-emnlp.555",
pages = "8278--8289",
abstract = "Non-autoregressive models have been widely used for various text generation tasks to accelerate the inference process but at the cost of generation quality to some extent. To achieve a good balance between inference speedup and generation quality, iterative NAR models like CMLM and Disco are proposed. Researchers have made much follow-up progress based on them, and some recent iterative models can achieve very promising performance while maintaining significant speedup. In this paper, we give more insights into iterative NAR models by exploring the anisotropic problem, i.e., the representations of distinct predicted target tokens are similar and indiscriminative. Upon the confirmation of the anisotropic problem in iterative NAR models, we first analyze the effectiveness of the contrastive learning method and further propose the Look Neighbors strategy to enhance the learning of token representations during training. Experiments on 4 WMT datasets show that our methods consistently improve the performance as well as alleviate the anisotropic problem of the conditional masked language model, even outperforming the current SoTA result on WMT14 EN $\rightarrow$ DE.",
}
| Non-autoregressive models have been widely used for various text generation tasks to accelerate the inference process but at the cost of generation quality to some extent. To achieve a good balance between inference speedup and generation quality, iterative NAR models like CMLM and Disco are proposed. Researchers have made much follow-up progress based on them, and some recent iterative models can achieve very promising performance while maintaining significant speedup. In this paper, we give more insights into iterative NAR models by exploring the anisotropic problem, i.e., the representations of distinct predicted target tokens are similar and indiscriminative. Upon the confirmation of the anisotropic problem in iterative NAR models, we first analyze the effectiveness of the contrastive learning method and further propose the Look Neighbors strategy to enhance the learning of token representations during training. Experiments on 4 WMT datasets show that our methods consistently improve the performance as well as alleviate the anisotropic problem of the conditional masked language model, even outperforming the current SoTA result on WMT14 EN $\rightarrow$ DE. | [
"Guo, Pei",
"Xiao, Yisheng",
"Li, Juntao",
"Ji, Yixin",
"Zhang, Min"
] | Isotropy-Enhanced Conditional Masked Language Models | findings-emnlp.555 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.556.bib | https://aclanthology.org/2023.findings-emnlp.556/ | @inproceedings{zhuocheng-etal-2023-scaling,
title = "Scaling Law for Document Neural Machine Translation",
author = "Zhuocheng, Zhang and
Gu, Shuhao and
Zhang, Min and
Feng, Yang",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.556",
doi = "10.18653/v1/2023.findings-emnlp.556",
pages = "8290--8303",
abstract = "The scaling laws of language models have played a significant role in advancing large language models. In order to promote the development of document translation, we systematically examine the scaling laws in this field. In this paper, we carry out an in-depth analysis of the influence of three factors on translation quality: model scale, data scale, and sequence length. Our findings reveal that increasing sequence length effectively enhances model performance when model size is limited. However, sequence length cannot be infinitely extended; it must be suitably aligned with the model scale and corpus volume. Further research shows that providing adequate context can effectively enhance the translation quality of a document{'}s initial portion. Nonetheless, exposure bias remains the primary factor hindering further improvement in translation quality for the latter half of the document.",
}
| The scaling laws of language models have played a significant role in advancing large language models. In order to promote the development of document translation, we systematically examine the scaling laws in this field. In this paper, we carry out an in-depth analysis of the influence of three factors on translation quality: model scale, data scale, and sequence length. Our findings reveal that increasing sequence length effectively enhances model performance when model size is limited. However, sequence length cannot be infinitely extended; it must be suitably aligned with the model scale and corpus volume. Further research shows that providing adequate context can effectively enhance the translation quality of a document{'}s initial portion. Nonetheless, exposure bias remains the primary factor hindering further improvement in translation quality for the latter half of the document. | [
"Zhuocheng, Zhang",
"Gu, Shuhao",
"Zhang, Min",
"Feng, Yang"
] | Scaling Law for Document Neural Machine Translation | findings-emnlp.556 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.557.bib | https://aclanthology.org/2023.findings-emnlp.557/ | @inproceedings{kheir-etal-2023-automatic,
title = "Automatic Pronunciation Assessment - A Review",
author = "Kheir, Yassine and
Ali, Ahmed and
Chowdhury, Shammur",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.557",
doi = "10.18653/v1/2023.findings-emnlp.557",
pages = "8304--8324",
abstract = "Pronunciation assessment and its application in computer-aided pronunciation training (CAPT) have seen impressive progress in recent years. With the rapid growth in language processing and deep learning over the past few years, there is a need for an updated review. In this paper, we review methods employed in pronunciation assessment for both phonemic and prosodic. We categorize the main challenges observed in prominent research trends, and highlight existing limitations, and available resources. This is followed by a discussion of the remaining challenges and possible directions for future work.",
}
| Pronunciation assessment and its application in computer-aided pronunciation training (CAPT) have seen impressive progress in recent years. With the rapid growth in language processing and deep learning over the past few years, there is a need for an updated review. In this paper, we review methods employed in pronunciation assessment for both phonemic and prosodic. We categorize the main challenges observed in prominent research trends, and highlight existing limitations, and available resources. This is followed by a discussion of the remaining challenges and possible directions for future work. | [
"Kheir, Yassine",
"Ali, Ahmed",
"Chowdhury, Shammur"
] | Automatic Pronunciation Assessment - A Review | findings-emnlp.557 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.558.bib | https://aclanthology.org/2023.findings-emnlp.558/ | @inproceedings{long-etal-2023-segmented,
title = "Segmented Recurrent Transformer: An Efficient Sequence-to-Sequence Model",
author = "Long, Yinghan and
Chowdhury, Sayeed and
Roy, Kaushik",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.558",
doi = "10.18653/v1/2023.findings-emnlp.558",
pages = "8325--8337",
abstract = "Transformers have shown dominant performance across a range of domains including language and vision. However, their computational cost grows quadratically with the sequence length, making their usage prohibitive for resource-constrained applications. To counter this, our approach is to divide the whole sequence into segments and apply attention to the individual segments. We propose a segmented recurrent transformer (SRformer) that combines segmented (local) attention with recurrent attention. The loss caused by reducing the attention window length is compensated by aggregating information across segments with recurrent attention. SRformer leverages Recurrent Accumulate-and-Fire (RAF) neurons{'} inherent memory to update the cumulative product of keys and values. The segmented attention and lightweight RAF neurons ensure the efficiency of the proposed transformer. Such an approach leads to models with sequential processing capability at a lower computation/memory cost. We apply the proposed method to T5 and BART transformers. The modified models are tested on summarization datasets including CNN-dailymail, XSUM, ArXiv, and MediaSUM. Notably, using segmented inputs of varied sizes, the proposed model achieves 6-22{\%} higher ROUGE1 scores than a segmented transformer and outperforms other recurrent transformer approaches. Furthermore, compared to full attention, the proposed model reduces the computational complexity of cross attention by around 40{\%}.",
}
| Transformers have shown dominant performance across a range of domains including language and vision. However, their computational cost grows quadratically with the sequence length, making their usage prohibitive for resource-constrained applications. To counter this, our approach is to divide the whole sequence into segments and apply attention to the individual segments. We propose a segmented recurrent transformer (SRformer) that combines segmented (local) attention with recurrent attention. The loss caused by reducing the attention window length is compensated by aggregating information across segments with recurrent attention. SRformer leverages Recurrent Accumulate-and-Fire (RAF) neurons{'} inherent memory to update the cumulative product of keys and values. The segmented attention and lightweight RAF neurons ensure the efficiency of the proposed transformer. Such an approach leads to models with sequential processing capability at a lower computation/memory cost. We apply the proposed method to T5 and BART transformers. The modified models are tested on summarization datasets including CNN-dailymail, XSUM, ArXiv, and MediaSUM. Notably, using segmented inputs of varied sizes, the proposed model achieves 6-22{\%} higher ROUGE1 scores than a segmented transformer and outperforms other recurrent transformer approaches. Furthermore, compared to full attention, the proposed model reduces the computational complexity of cross attention by around 40{\%}. | [
"Long, Yinghan",
"Chowdhury, Sayeed",
"Roy, Kaushik"
] | Segmented Recurrent Transformer: An Efficient Sequence-to-Sequence Model | findings-emnlp.558 | 2305.16340 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.559.bib | https://aclanthology.org/2023.findings-emnlp.559/ | @inproceedings{ma-etal-2023-punr,
title = "{PUNR}: Pre-training with User Behavior Modeling for News Recommendation",
author = "Ma, Guangyuan and
Liu, Hongtao and
W, Xing and
Qian, Wanhui and
Lv, Zhepeng and
Yang, Qing and
Hu, Songlin",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.559",
doi = "10.18653/v1/2023.findings-emnlp.559",
pages = "8338--8347",
abstract = "News recommendation aims to predict click behaviors based on user behaviors. How to effectively model the user representations is the key to recommending preferred news. Existing works are mostly focused on improvements in the supervised fine-tuning stage. However, there is still a lack of PLM-based unsupervised pre-training methods optimized for user representations. In this work, we propose an unsupervised pre-training paradigm with two tasks, i.e. user behavior masking and user behavior generation, both towards effective user behavior modeling. Firstly, we introduce the user behavior masking pre-training task to recover the masked user behaviors based on their contextual behaviors. In this way, the model could capture a much stronger and more comprehensive user news reading pattern. Besides, we incorporate a novel auxiliary user behavior generation pre-training task to enhance the user representation vector derived from the user encoder. We use the above pre-trained user modeling encoder to obtain news and user representations in downstream fine-tuning. Evaluations on the real-world news benchmark show significant performance improvements over existing baselines.",
}
| News recommendation aims to predict click behaviors based on user behaviors. How to effectively model the user representations is the key to recommending preferred news. Existing works are mostly focused on improvements in the supervised fine-tuning stage. However, there is still a lack of PLM-based unsupervised pre-training methods optimized for user representations. In this work, we propose an unsupervised pre-training paradigm with two tasks, i.e. user behavior masking and user behavior generation, both towards effective user behavior modeling. Firstly, we introduce the user behavior masking pre-training task to recover the masked user behaviors based on their contextual behaviors. In this way, the model could capture a much stronger and more comprehensive user news reading pattern. Besides, we incorporate a novel auxiliary user behavior generation pre-training task to enhance the user representation vector derived from the user encoder. We use the above pre-trained user modeling encoder to obtain news and user representations in downstream fine-tuning. Evaluations on the real-world news benchmark show significant performance improvements over existing baselines. | [
"Ma, Guangyuan",
"Liu, Hongtao",
"W, Xing",
"Qian, Wanhui",
"Lv, Zhepeng",
"Yang, Qing",
"Hu, Songlin"
] | PUNR: Pre-training with User Behavior Modeling for News Recommendation | findings-emnlp.559 | 2304.12633 | [
"https://github.com/ma787639046/punr"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.560.bib | https://aclanthology.org/2023.findings-emnlp.560/ | @inproceedings{sprueill-etal-2023-monte,
title = "{M}onte {C}arlo Thought Search: Large Language Model Querying for Complex Scientific Reasoning in Catalyst Design",
author = "Sprueill, Henry and
Edwards, Carl and
Olarte, Mariefel and
Sanyal, Udishnu and
Ji, Heng and
Choudhury, Sutanay",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.560",
doi = "10.18653/v1/2023.findings-emnlp.560",
pages = "8348--8365",
abstract = "Discovering novel catalysts requires complex reasoning involving multiple chemical properties and resultant trade-offs, leading to a combinatorial growth in the search space. While large language models (LLM) have demonstrated novel capabilities for chemistry through complex instruction following capabilities and high quality reasoning, a goal-driven combinatorial search using LLMs has not been explored in detail. In this work, we present a Monte Carlo Tree Search-based approach that improves beyond state-of-the-art chain-of-thought prompting variants to augment scientific reasoning. We introduce two new reasoning datasets: 1) a curation of computational chemistry simulations, and 2) diverse questions written by catalysis researchers for reasoning about novel chemical conversion processes. We improve over the best baseline by 25.8{\%} and find that our approach can augment scientist{'}s reasoning and discovery process with novel insights.",
}
| Discovering novel catalysts requires complex reasoning involving multiple chemical properties and resultant trade-offs, leading to a combinatorial growth in the search space. While large language models (LLM) have demonstrated novel capabilities for chemistry through complex instruction following capabilities and high quality reasoning, a goal-driven combinatorial search using LLMs has not been explored in detail. In this work, we present a Monte Carlo Tree Search-based approach that improves beyond state-of-the-art chain-of-thought prompting variants to augment scientific reasoning. We introduce two new reasoning datasets: 1) a curation of computational chemistry simulations, and 2) diverse questions written by catalysis researchers for reasoning about novel chemical conversion processes. We improve over the best baseline by 25.8{\%} and find that our approach can augment scientist{'}s reasoning and discovery process with novel insights. | [
"Sprueill, Henry",
"Edwards, Carl",
"Olarte, Mariefel",
"Sanyal, Udishnu",
"Ji, Heng",
"Choudhury, Sutanay"
] | Monte Carlo Thought Search: Large Language Model Querying for Complex Scientific Reasoning in Catalyst Design | findings-emnlp.560 | 2310.14420 | [
"https://github.com/pnnl/chemreasoner"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.561.bib | https://aclanthology.org/2023.findings-emnlp.561/ | @inproceedings{yan-etal-2023-measure,
title = "Measure Children{'}s Mindreading Ability with Machine Reading",
author = "Yan, Yuliang and
Wang, Xiaohua and
Zhou, Xiang and
Zheng, Xiaoqing and
Huang, Xuanjing",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.561",
doi = "10.18653/v1/2023.findings-emnlp.561",
pages = "8366--8375",
abstract = "Recently, much research in psychology has benefited from the advances in machine learning techniques. Some recent studies showed that it is possible to build automated scoring models for children{'}s mindreading. These models were trained on a set of manually-labeled question-response pairs, which were collected by asking children to answer one or two questions after a short story is told or a video clip is played. However, existing models did not take the features of the stories and video clips into account when scoring, which obviously will reduce the accuracy of the scoring models. Furthermore, considering that different psychological tests may contain the same questions, this approach cannot be extended to other related psychological test datasets. In this study, we proposed a multi-modal learning framework to leverage the features extracted from the stories and videos related to the questions being asked during the children{'}s mindreading evaluation. Experimental results show that the scores produced by the proposed models agree well with those graded by human experts, highlighting the potential of the proposed network architecture for practical automated children{'}s mindreading scoring systems.",
}
| Recently, much research in psychology has benefited from the advances in machine learning techniques. Some recent studies showed that it is possible to build automated scoring models for children{'}s mindreading. These models were trained on a set of manually-labeled question-response pairs, which were collected by asking children to answer one or two questions after a short story is told or a video clip is played. However, existing models did not take the features of the stories and video clips into account when scoring, which obviously will reduce the accuracy of the scoring models. Furthermore, considering that different psychological tests may contain the same questions, this approach cannot be extended to other related psychological test datasets. In this study, we proposed a multi-modal learning framework to leverage the features extracted from the stories and videos related to the questions being asked during the children{'}s mindreading evaluation. Experimental results show that the scores produced by the proposed models agree well with those graded by human experts, highlighting the potential of the proposed network architecture for practical automated children{'}s mindreading scoring systems. | [
"Yan, Yuliang",
"Wang, Xiaohua",
"Zhou, Xiang",
"Zheng, Xiaoqing",
"Huang, Xuanjing"
] | Measure Children's Mindreading Ability with Machine Reading | findings-emnlp.561 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.562.bib | https://aclanthology.org/2023.findings-emnlp.562/ | @inproceedings{liu-etal-2023-crosslingual-transfer,
title = "Crosslingual Transfer Learning for Low-Resource Languages Based on Multilingual Colexification Graphs",
author = "Liu, Yihong and
Ye, Haotian and
Weissweiler, Leonie and
Pei, Renhao and
Schuetze, Hinrich",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.562",
doi = "10.18653/v1/2023.findings-emnlp.562",
pages = "8376--8401",
abstract = "In comparative linguistics, colexification refers to the phenomenon of a lexical form conveying two or more distinct meanings. Existing work on colexification patterns relies on annotated word lists, limiting scalability and usefulness in NLP. In contrast, we identify colexification patterns of more than 2,000 concepts across 1,335 languages directly from an unannotated parallel corpus. We then propose simple and effective methods to build multilingual graphs from the colexification patterns: \textbf{ColexNet} and \textbf{ColexNet+}. ColexNet{'}s nodes are concepts and its edges are colexifications. In ColexNet+, concept nodes are additionally linked through intermediate nodes, each representing an ngram in one of 1,334 languages. We use ColexNet+ to train $\overrightarrow{\mbox{ColexNet+}}$, high-quality multilingual embeddings that are well-suited for transfer learning. In our experiments, we first show that ColexNet achieves high recall on CLICS, a dataset of crosslingual colexifications. We then evaluate $\overrightarrow{\mbox{ColexNet+}}$ on roundtrip translation, sentence retrieval and sentence classification and show that our embeddings surpass several transfer learning baselines. This demonstrates the benefits of using colexification as a source of information in multilingual NLP.",
}
| In comparative linguistics, colexification refers to the phenomenon of a lexical form conveying two or more distinct meanings. Existing work on colexification patterns relies on annotated word lists, limiting scalability and usefulness in NLP. In contrast, we identify colexification patterns of more than 2,000 concepts across 1,335 languages directly from an unannotated parallel corpus. We then propose simple and effective methods to build multilingual graphs from the colexification patterns: \textbf{ColexNet} and \textbf{ColexNet+}. ColexNet{'}s nodes are concepts and its edges are colexifications. In ColexNet+, concept nodes are additionally linked through intermediate nodes, each representing an ngram in one of 1,334 languages. We use ColexNet+ to train $\overrightarrow{\mbox{ColexNet+}}$, high-quality multilingual embeddings that are well-suited for transfer learning. In our experiments, we first show that ColexNet achieves high recall on CLICS, a dataset of crosslingual colexifications. We then evaluate $\overrightarrow{\mbox{ColexNet+}}$ on roundtrip translation, sentence retrieval and sentence classification and show that our embeddings surpass several transfer learning baselines. This demonstrates the benefits of using colexification as a source of information in multilingual NLP. | [
"Liu, Yihong",
"Ye, Haotian",
"Weissweiler, Leonie",
"Pei, Renhao",
"Schuetze, Hinrich"
] | Crosslingual Transfer Learning for Low-Resource Languages Based on Multilingual Colexification Graphs | findings-emnlp.562 | 2305.12818 | [
"https://github.com/yihongl1u/colexificationnet"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.563.bib | https://aclanthology.org/2023.findings-emnlp.563/ | @inproceedings{papadimitriou-jurafsky-2023-injecting,
title = "Injecting structural hints: Using language models to study inductive biases in language learning",
author = "Papadimitriou, Isabel and
Jurafsky, Dan",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.563",
doi = "10.18653/v1/2023.findings-emnlp.563",
pages = "8402--8413",
abstract = "Both humans and transformer language models are able to learn language without explicit structural supervision. What cognitive inductive biases make this learning possible? Here, we examine the effect of different inductive learning biases by actively controlling the inductive biases of artificial learners: we structurally bias models by pretraining on synthetic formally-structured data, and evaluate these structural biases by fine-tuning on three typologically-distant human languages: English, Japanese, and Basque. We investigate the effect on downstream language perplexity of three types of inductive bias: 1) recursive, hierarchical processing 2) unrestricted token-token dependencies that can{'}t be modeled by context-free grammars, and 3) a Zipfian power-law vocabulary distribution. We show that complex, non-context-free interactions between tokens form the best inductive biases. Our study leverages the capabilities of transformer models to run controlled language learning experiments that are not possible to run on humans, and surfaces hypotheses about the structures that facilitate language learning in both humans and machines.",
}
| Both humans and transformer language models are able to learn language without explicit structural supervision. What cognitive inductive biases make this learning possible? Here, we examine the effect of different inductive learning biases by actively controlling the inductive biases of artificial learners: we structurally bias models by pretraining on synthetic formally-structured data, and evaluate these structural biases by fine-tuning on three typologically-distant human languages: English, Japanese, and Basque. We investigate the effect on downstream language perplexity of three types of inductive bias: 1) recursive, hierarchical processing 2) unrestricted token-token dependencies that can{'}t be modeled by context-free grammars, and 3) a Zipfian power-law vocabulary distribution. We show that complex, non-context-free interactions between tokens form the best inductive biases. Our study leverages the capabilities of transformer models to run controlled language learning experiments that are not possible to run on humans, and surfaces hypotheses about the structures that facilitate language learning in both humans and machines. | [
"Papadimitriou, Isabel",
"Jurafsky, Dan"
] | Injecting structural hints: Using language models to study inductive biases in language learning | findings-emnlp.563 | 2304.13060 | [
"https://github.com/toizzy/injecting-structural-hints"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.564.bib | https://aclanthology.org/2023.findings-emnlp.564/ | @inproceedings{thai-etal-2023-machine,
title = "Machine Reading Comprehension using Case-based Reasoning",
author = "Thai, Dung and
Agarwal, Dhruv and
Chaudhary, Mudit and
Zhao, Wenlong and
Das, Rajarshi and
Lee, Jay-Yoon and
Hajishirzi, Hannaneh and
Zaheer, Manzil and
McCallum, Andrew",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.564",
doi = "10.18653/v1/2023.findings-emnlp.564",
pages = "8414--8428",
abstract = "We present an accurate and interpretable method for answer extraction in machine reading comprehension that is reminiscent of case-based reasoning (CBR) from classical AI. Our method (CBR-MRC) builds upon the hypothesis that contextualized answers to similar questions share semantic similarities with each other. Given a test question, CBR-MRC first retrieves a set of similar cases from a nonparametric memory and then predicts an answer by selecting the span in the test context that is most similar to the contextualized representations of answers in the retrieved cases. The semi-parametric nature of our approach allows it to attribute a prediction to the specific set of evidence cases, making it a desirable choice for building reliable and debuggable QA systems. We show that CBR-MRC provides high accuracy comparable with large reader models and outperforms baselines by 11.5 and 8.4 EM on NaturalQuestions and NewsQA, respectively. Further, we demonstrate the ability of CBR-MRC in identifying not just the correct answer tokens but also the span with the most relevant supporting evidence. Lastly, we observe that contexts for certain question types show higher lexical diversity than others and find that CBR-MRC is robust to these variations while performance using fully-parametric methods drops.",
}
| We present an accurate and interpretable method for answer extraction in machine reading comprehension that is reminiscent of case-based reasoning (CBR) from classical AI. Our method (CBR-MRC) builds upon the hypothesis that contextualized answers to similar questions share semantic similarities with each other. Given a test question, CBR-MRC first retrieves a set of similar cases from a nonparametric memory and then predicts an answer by selecting the span in the test context that is most similar to the contextualized representations of answers in the retrieved cases. The semi-parametric nature of our approach allows it to attribute a prediction to the specific set of evidence cases, making it a desirable choice for building reliable and debuggable QA systems. We show that CBR-MRC provides high accuracy comparable with large reader models and outperforms baselines by 11.5 and 8.4 EM on NaturalQuestions and NewsQA, respectively. Further, we demonstrate the ability of CBR-MRC in identifying not just the correct answer tokens but also the span with the most relevant supporting evidence. Lastly, we observe that contexts for certain question types show higher lexical diversity than others and find that CBR-MRC is robust to these variations while performance using fully-parametric methods drops. | [
"Thai, Dung",
"Agarwal, Dhruv",
"Chaudhary, Mudit",
"Zhao, Wenlong",
"Das, Rajarshi",
"Lee, Jay-Yoon",
"Hajishirzi, Hannaneh",
"Zaheer, Manzil",
"McCallum, Andrew"
] | Machine Reading Comprehension using Case-based Reasoning | findings-emnlp.564 | 2305.14815 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.565.bib | https://aclanthology.org/2023.findings-emnlp.565/ | @inproceedings{kuang-etal-2023-unleashing,
title = "Unleashing the Power of Language Models in Text-Attributed Graph",
author = "Kuang, Haoyu and
Xu, Jiarong and
Zhang, Haozhe and
Zhao, Zuyu and
Zhang, Qi and
Huang, Xuanjing and
Wei, Zhongyu",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.565",
doi = "10.18653/v1/2023.findings-emnlp.565",
pages = "8429--8441",
abstract = "Representation learning on graph has been demonstrated to be a powerful tool for solving real-world problems. Text-attributed graph carries both semantic and structural information among different types of graphs. Existing works have paved the way for knowledge extraction of this type of data by leveraging language models or graph neural networks or combination of them. However, these works suffer from issues like underutilization of relationships between nodes or words or unaffordable memory cost. In this paper, we propose a Node Representation Update Pre-training Architecture based on Co-modeling Text and Graph (NRUP). In NRUP, we construct a hierarchical text-attributed graph that incorporates both original nodes and word nodes. Meanwhile, we apply four self-supervised tasks for different level of constructed graph. We further design the pre-training framework to update the features of nodes during training epochs. We conduct the experiment on the benchmark dataset ogbn-arxiv. Our method achieves outperformance compared to baselines, fully demonstrating its validity and generalization.",
}
| Representation learning on graph has been demonstrated to be a powerful tool for solving real-world problems. Text-attributed graph carries both semantic and structural information among different types of graphs. Existing works have paved the way for knowledge extraction of this type of data by leveraging language models or graph neural networks or combination of them. However, these works suffer from issues like underutilization of relationships between nodes or words or unaffordable memory cost. In this paper, we propose a Node Representation Update Pre-training Architecture based on Co-modeling Text and Graph (NRUP). In NRUP, we construct a hierarchical text-attributed graph that incorporates both original nodes and word nodes. Meanwhile, we apply four self-supervised tasks for different level of constructed graph. We further design the pre-training framework to update the features of nodes during training epochs. We conduct the experiment on the benchmark dataset ogbn-arxiv. Our method achieves outperformance compared to baselines, fully demonstrating its validity and generalization. | [
"Kuang, Haoyu",
"Xu, Jiarong",
"Zhang, Haozhe",
"Zhao, Zuyu",
"Zhang, Qi",
"Huang, Xuanjing",
"Wei, Zhongyu"
] | Unleashing the Power of Language Models in Text-Attributed Graph | findings-emnlp.565 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.566.bib | https://aclanthology.org/2023.findings-emnlp.566/ | @inproceedings{utpala-etal-2023-locally,
title = "Locally Differentially Private Document Generation Using Zero Shot Prompting",
author = "Utpala, Saiteja and
Hooker, Sara and
Chen, Pin-Yu",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.566",
doi = "10.18653/v1/2023.findings-emnlp.566",
pages = "8442--8457",
abstract = "Numerous studies have highlighted the privacy risks associated with large language models. Our research offers a unique perspective by demonstrating that pretrained large language models can effectively contribute to privacy preservation. We propose a locally differentially private mechanism called DP-Prompt, which leverages the power of pretrained large language models and zero-shot prompting to counter author de-anonymization attacks while minimizing the impact on downstream utility. When DP-Prompt is used with a powerful language model like ChatGPT (gpt-3.5), we observe a notable reduction in the success rate of de-anonymization attacks, showing that it surpasses existing approaches by a considerable margin despite its simpler design. For instance, in the case of the IMDB dataset, DP-Prompt (with ChatGPT) perfectly recovers the clean sentiment F1 score while achieving a 46{\%} reduction in author identification F1 score against static attackers and a 26{\%} reduction against adaptive attackers. We conduct extensive experiments across six open-source large language models, ranging up to 7 billion parameters, to analyze various effects of the privacy-utility tradeoff.",
}
| Numerous studies have highlighted the privacy risks associated with large language models. Our research offers a unique perspective by demonstrating that pretrained large language models can effectively contribute to privacy preservation. We propose a locally differentially private mechanism called DP-Prompt, which leverages the power of pretrained large language models and zero-shot prompting to counter author de-anonymization attacks while minimizing the impact on downstream utility. When DP-Prompt is used with a powerful language model like ChatGPT (gpt-3.5), we observe a notable reduction in the success rate of de-anonymization attacks, showing that it surpasses existing approaches by a considerable margin despite its simpler design. For instance, in the case of the IMDB dataset, DP-Prompt (with ChatGPT) perfectly recovers the clean sentiment F1 score while achieving a 46{\%} reduction in author identification F1 score against static attackers and a 26{\%} reduction against adaptive attackers. We conduct extensive experiments across six open-source large language models, ranging up to 7 billion parameters, to analyze various effects of the privacy-utility tradeoff. | [
"Utpala, Saiteja",
"Hooker, Sara",
"Chen, Pin-Yu"
] | Locally Differentially Private Document Generation Using Zero Shot Prompting | findings-emnlp.566 | 2310.16111 | [
"https://github.com/saitejautpala/dp_prompt"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.567.bib | https://aclanthology.org/2023.findings-emnlp.567/ | @inproceedings{ghose-poupart-2023-contrastive,
title = "Contrastive Deterministic Autoencoders For Language Modeling",
author = "Ghose, Amur and
Poupart, Pascal",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.567",
doi = "10.18653/v1/2023.findings-emnlp.567",
pages = "8458--8476",
abstract = "Variational autoencoders (VAEs) are a popular family of generative models with wide applicability. Training VAEs, especially for text, often runs into the issue of posterior collapse, resulting in loss of representation quality. Deterministic autoencoders avoid this issue, and have been explored particularly well for images. It is however unclear how to best modify a deterministic model designed for images into a successful one for text. We show that with suitable adaptations, we can significantly improve on batch-normed VAEs (BN-VAEs), a strong benchmark for language modeling with VAEs, by replacing them with analogous deterministic models. We employ techniques from contrastive learning to control the entropy of the aggregate posterior of these models to make it Gaussian. The resulting models skip reparametrization steps in VAE modeling and avoid posterior collapse, while outperforming a broad range of VAE models on text generation and downstream tasks from representations. These improvements are shown to be consistent across both LSTM and Transformer-based VAE architectures. Appropriate comparisons to BERT/GPT-2 based results are also included. We also qualitatively examine the latent space through interpolation to supplement the quantitative aspects of the model.",
}
| Variational autoencoders (VAEs) are a popular family of generative models with wide applicability. Training VAEs, especially for text, often runs into the issue of posterior collapse, resulting in loss of representation quality. Deterministic autoencoders avoid this issue, and have been explored particularly well for images. It is however unclear how to best modify a deterministic model designed for images into a successful one for text. We show that with suitable adaptations, we can significantly improve on batch-normed VAEs (BN-VAEs), a strong benchmark for language modeling with VAEs, by replacing them with analogous deterministic models. We employ techniques from contrastive learning to control the entropy of the aggregate posterior of these models to make it Gaussian. The resulting models skip reparametrization steps in VAE modeling and avoid posterior collapse, while outperforming a broad range of VAE models on text generation and downstream tasks from representations. These improvements are shown to be consistent across both LSTM and Transformer-based VAE architectures. Appropriate comparisons to BERT/GPT-2 based results are also included. We also qualitatively examine the latent space through interpolation to supplement the quantitative aspects of the model. | [
"Ghose, Amur",
"Poupart, Pascal"
] | Contrastive Deterministic Autoencoders For Language Modeling | findings-emnlp.567 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.568.bib | https://aclanthology.org/2023.findings-emnlp.568/ | @inproceedings{mcinerney-etal-2023-chill,
title = "{CH}i{LL}: Zero-shot Custom Interpretable Feature Extraction from Clinical Notes with Large Language Models",
author = "McInerney, Denis and
Young, Geoffrey and
van de Meent, Jan-Willem and
Wallace, Byron",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.568",
doi = "10.18653/v1/2023.findings-emnlp.568",
pages = "8477--8494",
abstract = "We propose CHiLL (Crafting High-Level Latents), an approach for natural-language specification of features for linear models. CHiLL prompts LLMs with expert-crafted queries to generate interpretable features from health records. The resulting noisy labels are then used to train a simple linear classifier. Generating features based on queries to an LLM can empower physicians to use their domain expertise to craft features that are clinically meaningful for a downstream task of interest, without having to manually extract these from raw EHR. We are motivated by a real-world risk prediction task, but as a reproducible proxy, we use MIMIC-III and MIMIC-CXR data and standard predictive tasks (e.g., 30-day readmission) to evaluate this approach. We find that linear models using automatically extracted features are comparably performant to models using reference features, and provide greater interpretability than linear models using {``}Bag-of-Words{''} features. We verify that learned feature weights align well with clinical expectations.",
}
| We propose CHiLL (Crafting High-Level Latents), an approach for natural-language specification of features for linear models. CHiLL prompts LLMs with expert-crafted queries to generate interpretable features from health records. The resulting noisy labels are then used to train a simple linear classifier. Generating features based on queries to an LLM can empower physicians to use their domain expertise to craft features that are clinically meaningful for a downstream task of interest, without having to manually extract these from raw EHR. We are motivated by a real-world risk prediction task, but as a reproducible proxy, we use MIMIC-III and MIMIC-CXR data and standard predictive tasks (e.g., 30-day readmission) to evaluate this approach. We find that linear models using automatically extracted features are comparably performant to models using reference features, and provide greater interpretability than linear models using {``}Bag-of-Words{''} features. We verify that learned feature weights align well with clinical expectations. | [
"McInerney, Denis",
"Young, Geoffrey",
"van de Meent, Jan-Willem",
"Wallace, Byron"
] | CHiLL: Zero-shot Custom Interpretable Feature Extraction from Clinical Notes with Large Language Models | findings-emnlp.568 | 2302.12343 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.569.bib | https://aclanthology.org/2023.findings-emnlp.569/ | @inproceedings{levy-etal-2023-guiding,
title = "Guiding {LLM} to Fool Itself: Automatically Manipulating Machine Reading Comprehension Shortcut Triggers",
author = "Levy, Mosh and
Ravfogel, Shauli and
Goldberg, Yoav",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.569",
doi = "10.18653/v1/2023.findings-emnlp.569",
pages = "8495--8505",
abstract = "Recent applications of LLMs in Machine Reading Comprehension (MRC) systems have shown impressive results, but the use of shortcuts, mechanisms triggered by features spuriously correlated to the true label, has emerged as a potential threat to their reliability. We analyze the problem from two angles: LLMs as editors, guided to edit text to mislead LLMs; and LLMs as readers, who answer questions based on the edited text. We introduce a framework that guides an editor to add potential shortcuts-triggers to samples. Using GPT4 as the editor, we find it can successfully edit trigger shortcut in samples that fool LLMs. Analysing LLMs as readers, we observe that even capable LLMs can be deceived using shortcut knowledge. Strikingly, we discover that GPT4 can be deceived by its own edits (15{\%} drop in F1). Our findings highlight inherent vulnerabilities of LLMs to shortcut manipulations. We publish ShortcutQA, a curated dataset generated by our framework for future research.",
}
| Recent applications of LLMs in Machine Reading Comprehension (MRC) systems have shown impressive results, but the use of shortcuts, mechanisms triggered by features spuriously correlated to the true label, has emerged as a potential threat to their reliability. We analyze the problem from two angles: LLMs as editors, guided to edit text to mislead LLMs; and LLMs as readers, who answer questions based on the edited text. We introduce a framework that guides an editor to add potential shortcuts-triggers to samples. Using GPT4 as the editor, we find it can successfully edit trigger shortcut in samples that fool LLMs. Analysing LLMs as readers, we observe that even capable LLMs can be deceived using shortcut knowledge. Strikingly, we discover that GPT4 can be deceived by its own edits (15{\%} drop in F1). Our findings highlight inherent vulnerabilities of LLMs to shortcut manipulations. We publish ShortcutQA, a curated dataset generated by our framework for future research. | [
"Levy, Mosh",
"Ravfogel, Shauli",
"Goldberg, Yoav"
] | Guiding LLM to Fool Itself: Automatically Manipulating Machine Reading Comprehension Shortcut Triggers | findings-emnlp.569 | 2310.18360 | [
"https://github.com/mosh0110/guiding-llm"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.570.bib | https://aclanthology.org/2023.findings-emnlp.570/ | @inproceedings{chen-etal-2023-large,
title = "Large Language Models Meet Harry Potter: A Dataset for Aligning Dialogue Agents with Characters",
author = "Chen, Nuo and
Wang, Yan and
Jiang, Haiyun and
Cai, Deng and
Li, Yuhan and
Chen, Ziyang and
Wang, Longyue and
Li, Jia",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.570",
doi = "10.18653/v1/2023.findings-emnlp.570",
pages = "8506--8520",
abstract = "In recent years, Dialogue-style Large Language Models (LLMs) such as ChatGPT and GPT4 have demonstrated immense potential in constructing open-domain dialogue agents. However, aligning these agents with specific characters or individuals remains a considerable challenge due to the complexities of character representation and the lack of comprehensive annotations. In this paper, we introduce the Harry Potter Dialogue (HPD) dataset, designed to advance the study of dialogue agents and character alignment. The dataset encompasses all dialogue sessions (in both English and Chinese) from the Harry Potter series and is annotated with vital background information, including dialogue scenes, speakers, character relationships, and attributes. These extensive annotations may empower LLMs to unlock character-driven dialogue capabilities. Furthermore, it can serve as a universal benchmark for evaluating how well can a LLM aligning with a specific character. We benchmark LLMs on HPD using both fine-tuning and in-context learning settings. Evaluation results reveal that although there is substantial room for improvement in generating high-quality, character-aligned responses, the proposed dataset is valuable in guiding models toward responses that better align with the character of Harry Potter.",
}
| In recent years, Dialogue-style Large Language Models (LLMs) such as ChatGPT and GPT4 have demonstrated immense potential in constructing open-domain dialogue agents. However, aligning these agents with specific characters or individuals remains a considerable challenge due to the complexities of character representation and the lack of comprehensive annotations. In this paper, we introduce the Harry Potter Dialogue (HPD) dataset, designed to advance the study of dialogue agents and character alignment. The dataset encompasses all dialogue sessions (in both English and Chinese) from the Harry Potter series and is annotated with vital background information, including dialogue scenes, speakers, character relationships, and attributes. These extensive annotations may empower LLMs to unlock character-driven dialogue capabilities. Furthermore, it can serve as a universal benchmark for evaluating how well can a LLM aligning with a specific character. We benchmark LLMs on HPD using both fine-tuning and in-context learning settings. Evaluation results reveal that although there is substantial room for improvement in generating high-quality, character-aligned responses, the proposed dataset is valuable in guiding models toward responses that better align with the character of Harry Potter. | [
"Chen, Nuo",
"Wang, Yan",
"Jiang, Haiyun",
"Cai, Deng",
"Li, Yuhan",
"Chen, Ziyang",
"Wang, Longyue",
"Li, Jia"
] | Large Language Models Meet Harry Potter: A Dataset for Aligning Dialogue Agents with Characters | findings-emnlp.570 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.571.bib | https://aclanthology.org/2023.findings-emnlp.571/ | @inproceedings{brimacombe-zhou-2023-quick,
title = "Quick Back-Translation for Unsupervised Machine Translation",
author = "Brimacombe, Benjamin and
Zhou, Jiawei",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.571",
doi = "10.18653/v1/2023.findings-emnlp.571",
pages = "8521--8534",
abstract = "The field of unsupervised machine translation has seen significant advancement from the marriage of the Transformer and the back-translation algorithm. The Transformer is a powerful generative model, and back-translation leverages Transformer{'}s high-quality translations for iterative self-improvement. However, the Transformer is encumbered by the run-time of autoregressive inference during back-translation, and back-translation is limited by a lack of synthetic data efficiency. We propose a two-for-one improvement to Transformer back-translation: Quick Back-Translation (QBT). QBT re-purposes the encoder as a generative model, and uses encoder-generated sequences to train the decoder in conjunction with the original autoregressive back-translation step, improving data throughput and utilization. Experiments on various WMT benchmarks demonstrate that a relatively small number of refining steps of QBT improve current unsupervised machine translation models, and that QBT dramatically outperforms standard back-translation only method in terms of training efficiency for comparable translation qualities.",
}
| The field of unsupervised machine translation has seen significant advancement from the marriage of the Transformer and the back-translation algorithm. The Transformer is a powerful generative model, and back-translation leverages Transformer{'}s high-quality translations for iterative self-improvement. However, the Transformer is encumbered by the run-time of autoregressive inference during back-translation, and back-translation is limited by a lack of synthetic data efficiency. We propose a two-for-one improvement to Transformer back-translation: Quick Back-Translation (QBT). QBT re-purposes the encoder as a generative model, and uses encoder-generated sequences to train the decoder in conjunction with the original autoregressive back-translation step, improving data throughput and utilization. Experiments on various WMT benchmarks demonstrate that a relatively small number of refining steps of QBT improve current unsupervised machine translation models, and that QBT dramatically outperforms standard back-translation only method in terms of training efficiency for comparable translation qualities. | [
"Brimacombe, Benjamin",
"Zhou, Jiawei"
] | Quick Back-Translation for Unsupervised Machine Translation | findings-emnlp.571 | 2312.00912 | [
"https://github.com/bbrimacombe/quick-back-translation"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.572.bib | https://aclanthology.org/2023.findings-emnlp.572/ | @inproceedings{cho-etal-2023-sir,
title = "{SIR}-{ABSC}: Incorporating Syntax into {R}o{BERT}a-based Sentiment Analysis Models with a Special Aggregator Token",
author = "Cho, Ikhyun and
Jung, Yoonhwa and
Hockenmaier, Julia",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.572",
doi = "10.18653/v1/2023.findings-emnlp.572",
pages = "8535--8550",
abstract = "We present a simple, but effective method to incorporate syntactic dependency information directly into transformer-based language models (e.g. RoBERTa) for tasks such as Aspect-Based Sentiment Classification (ABSC), where the desired output depends on specific input tokens. In contrast to prior approaches to ABSC that capture syntax by combining language models with graph neural networks over dependency trees, our model, Syntax-Integrated RoBERTa for ABSC (SIR-ABSC) incorporates syntax directly into the language model by using a novel aggregator token. Yet, SIR-ABSC outperforms these more complex models, yielding new state-of-the-art results on ABSC.",
}
| We present a simple, but effective method to incorporate syntactic dependency information directly into transformer-based language models (e.g. RoBERTa) for tasks such as Aspect-Based Sentiment Classification (ABSC), where the desired output depends on specific input tokens. In contrast to prior approaches to ABSC that capture syntax by combining language models with graph neural networks over dependency trees, our model, Syntax-Integrated RoBERTa for ABSC (SIR-ABSC) incorporates syntax directly into the language model by using a novel aggregator token. Yet, SIR-ABSC outperforms these more complex models, yielding new state-of-the-art results on ABSC. | [
"Cho, Ikhyun",
"Jung, Yoonhwa",
"Hockenmaier, Julia"
] | SIR-ABSC: Incorporating Syntax into RoBERTa-based Sentiment Analysis Models with a Special Aggregator Token | findings-emnlp.572 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.573.bib | https://aclanthology.org/2023.findings-emnlp.573/ | @inproceedings{syed-etal-2023-citance,
title = "Citance-Contextualized Summarization of Scientific Papers",
author = "Syed, Shahbaz and
Hakimi, Ahmad and
Al-Khatib, Khalid and
Potthast, Martin",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.573",
doi = "10.18653/v1/2023.findings-emnlp.573",
pages = "8551--8568",
abstract = "Current approaches to automatic summarization of scientific papers generate informative summaries in the form of abstracts. However, abstracts are not intended to show the relationship between a paper and the references cited in it. We propose a new contextualized summarization approach that can generate an informative summary conditioned on a given sentence containing the citation of a reference (a so-called {``}citance{''}). This summary outlines content of the cited paper relevant to the citation location. Thus, our approach extracts and models the citances of a paper, retrieves relevant passages from cited papers, and generates abstractive summaries tailored to each citance. We evaluate our approach using **Webis-Context-SciSumm-2023**, a new dataset containing 540K computer science papers and 4.6M citances therein.",
}
| Current approaches to automatic summarization of scientific papers generate informative summaries in the form of abstracts. However, abstracts are not intended to show the relationship between a paper and the references cited in it. We propose a new contextualized summarization approach that can generate an informative summary conditioned on a given sentence containing the citation of a reference (a so-called {``}citance{''}). This summary outlines content of the cited paper relevant to the citation location. Thus, our approach extracts and models the citances of a paper, retrieves relevant passages from cited papers, and generates abstractive summaries tailored to each citance. We evaluate our approach using **Webis-Context-SciSumm-2023**, a new dataset containing 540K computer science papers and 4.6M citances therein. | [
"Syed, Shahbaz",
"Hakimi, Ahmad",
"Al-Khatib, Khalid",
"Potthast, Martin"
] | Citance-Contextualized Summarization of Scientific Papers | findings-emnlp.573 | 2311.02408 | [
"https://github.com/webis-de/emnlp-23"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.574.bib | https://aclanthology.org/2023.findings-emnlp.574/ | @inproceedings{tsiamas-etal-2023-segaugment,
title = "{S}eg{A}ugment: Maximizing the Utility of Speech Translation Data with Segmentation-based Augmentations",
author = "Tsiamas, Ioannis and
Fonollosa, Jos{\'e} and
Costa-juss{\`a}, Marta",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.574",
doi = "10.18653/v1/2023.findings-emnlp.574",
pages = "8569--8588",
abstract = "End-to-end Speech Translation is hindered by a lack of available data resources. While most of them are based on documents, a sentence-level version is available, which is however single and static, potentially impeding the usefulness of the data. We propose a new data augmentation strategy, SegAugment, to address this issue by generating multiple alternative sentence-level versions of a dataset. Our method utilizes an Audio Segmentation system, which re-segments the speech of each document with different length constraints, after which we obtain the target text via alignment methods. Experiments demonstrate consistent gains across eight language pairs in MuST-C, with an average increase of 2.5 BLEU points, and up to 5 BLEU for low-resource scenarios in mTEDx. Furthermore, when combined with a strong system, SegAugment obtains state-of-the-art results in MuST-C. Finally, we show that the proposed method can also successfully augment sentence-level datasets, and that it enables Speech Translation models to close the gap between the manual and automatic segmentation at inference time.",
}
| End-to-end Speech Translation is hindered by a lack of available data resources. While most of them are based on documents, a sentence-level version is available, which is however single and static, potentially impeding the usefulness of the data. We propose a new data augmentation strategy, SegAugment, to address this issue by generating multiple alternative sentence-level versions of a dataset. Our method utilizes an Audio Segmentation system, which re-segments the speech of each document with different length constraints, after which we obtain the target text via alignment methods. Experiments demonstrate consistent gains across eight language pairs in MuST-C, with an average increase of 2.5 BLEU points, and up to 5 BLEU for low-resource scenarios in mTEDx. Furthermore, when combined with a strong system, SegAugment obtains state-of-the-art results in MuST-C. Finally, we show that the proposed method can also successfully augment sentence-level datasets, and that it enables Speech Translation models to close the gap between the manual and automatic segmentation at inference time. | [
"Tsiamas, Ioannis",
"Fonollosa, Jos{\\'e}",
"Costa-juss{\\`a}, Marta"
] | SegAugment: Maximizing the Utility of Speech Translation Data with Segmentation-based Augmentations | findings-emnlp.574 | 2212.09699 | [
"https://github.com/mt-upc/SegAugment"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.575.bib | https://aclanthology.org/2023.findings-emnlp.575/ | @inproceedings{ma-etal-2023-intersectional,
title = "Intersectional Stereotypes in Large Language Models: Dataset and Analysis",
author = "Ma, Weicheng and
Chiang, Brian and
Wu, Tong and
Wang, Lili and
Vosoughi, Soroush",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.575",
doi = "10.18653/v1/2023.findings-emnlp.575",
pages = "8589--8597",
abstract = "Despite many stereotypes targeting intersectional demographic groups, prior studies on stereotypes within Large Language Models (LLMs) primarily focus on broader, individual categories. This research bridges this gap by introducing a novel dataset of intersectional stereotypes, curated with the assistance of the ChatGPT model and manually validated. Moreover, this paper offers a comprehensive analysis of intersectional stereotype propagation in three contemporary LLMs by leveraging this dataset. The findings underscore the urgency of focusing on intersectional biases in ongoing efforts to reduce stereotype prevalence in LLMs.",
}
| Despite many stereotypes targeting intersectional demographic groups, prior studies on stereotypes within Large Language Models (LLMs) primarily focus on broader, individual categories. This research bridges this gap by introducing a novel dataset of intersectional stereotypes, curated with the assistance of the ChatGPT model and manually validated. Moreover, this paper offers a comprehensive analysis of intersectional stereotype propagation in three contemporary LLMs by leveraging this dataset. The findings underscore the urgency of focusing on intersectional biases in ongoing efforts to reduce stereotype prevalence in LLMs. | [
"Ma, Weicheng",
"Chiang, Brian",
"Wu, Tong",
"Wang, Lili",
"Vosoughi, Soroush"
] | Intersectional Stereotypes in Large Language Models: Dataset and Analysis | findings-emnlp.575 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.576.bib | https://aclanthology.org/2023.findings-emnlp.576/ | @inproceedings{wang-etal-2023-dataset,
title = "Dataset Bias Mitigation in Multiple-Choice Visual Question Answering and Beyond",
author = "Wang, Zhecan and
Chen, Long and
You, Haoxuan and
Xu, Keyang and
He, Yicheng and
Li, Wenhao and
Codella, Noel and
Chang, Kai-Wei and
Chang, Shih-Fu",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.576",
doi = "10.18653/v1/2023.findings-emnlp.576",
pages = "8598--8617",
abstract = "Vision-language (VL) understanding tasks evaluate models{'} comprehension of complex visual scenes through multiple-choice questions. However, we have identified two dataset biases that models can exploit as shortcuts to resolve various VL tasks correctly without proper understanding. The first type of dataset bias is Unbalanced Matching bias, where the correct answer overlaps the question and image more than the incorrect answers. The second type of dataset bias is Distractor Similarity bias, where incorrect answers are overly dissimilar to the correct answer but significantly similar to other incorrect answers within the same sample. To address these dataset biases, we first propose Adversarial Data Synthesis (ADS) to generate synthetic training and debiased evaluation data. We then introduce Intra-sample Counterfactual Training (ICT) to assist models in utilizing the synthesized training data, particularly the counterfactual data, via focusing on intra-sample differentiation. Extensive experiments demonstrate the effectiveness of ADS and ICT in consistently improving model performance across different benchmarks, even in domain-shifted scenarios.",
}
| Vision-language (VL) understanding tasks evaluate models{'} comprehension of complex visual scenes through multiple-choice questions. However, we have identified two dataset biases that models can exploit as shortcuts to resolve various VL tasks correctly without proper understanding. The first type of dataset bias is Unbalanced Matching bias, where the correct answer overlaps the question and image more than the incorrect answers. The second type of dataset bias is Distractor Similarity bias, where incorrect answers are overly dissimilar to the correct answer but significantly similar to other incorrect answers within the same sample. To address these dataset biases, we first propose Adversarial Data Synthesis (ADS) to generate synthetic training and debiased evaluation data. We then introduce Intra-sample Counterfactual Training (ICT) to assist models in utilizing the synthesized training data, particularly the counterfactual data, via focusing on intra-sample differentiation. Extensive experiments demonstrate the effectiveness of ADS and ICT in consistently improving model performance across different benchmarks, even in domain-shifted scenarios. | [
"Wang, Zhecan",
"Chen, Long",
"You, Haoxuan",
"Xu, Keyang",
"He, Yicheng",
"Li, Wenhao",
"Codella, Noel",
"Chang, Kai-Wei",
"Chang, Shih-Fu"
] | Dataset Bias Mitigation in Multiple-Choice Visual Question Answering and Beyond | findings-emnlp.576 | 2310.14670 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.577.bib | https://aclanthology.org/2023.findings-emnlp.577/ | @inproceedings{schlichtkrull-etal-2023-intended,
title = "The Intended Uses of Automated Fact-Checking Artefacts: Why, How and Who",
author = "Schlichtkrull, Michael and
Ousidhoum, Nedjma and
Vlachos, Andreas",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.577",
doi = "10.18653/v1/2023.findings-emnlp.577",
pages = "8618--8642",
abstract = "Automated fact-checking is often presented as an epistemic tool that fact-checkers, social media consumers, and other stakeholders can use to fight misinformation. Nevertheless, few papers thoroughly discuss \textit{how}. We document this by analysing 100 highly-cited papers, and annotating epistemic elements related to intended use, i.e., means, ends, and stakeholders. We find that narratives leaving out some of these aspects are common, that many papers propose inconsistent means and ends, and that the feasibility of suggested strategies rarely has empirical backing. We argue that this vagueness actively hinders the technology from reaching its goals, as it encourages overclaiming, limits criticism, and prevents stakeholder feedback. Accordingly, we provide several recommendations for thinking and writing about the use of fact-checking artefacts.",
}
| Automated fact-checking is often presented as an epistemic tool that fact-checkers, social media consumers, and other stakeholders can use to fight misinformation. Nevertheless, few papers thoroughly discuss \textit{how}. We document this by analysing 100 highly-cited papers, and annotating epistemic elements related to intended use, i.e., means, ends, and stakeholders. We find that narratives leaving out some of these aspects are common, that many papers propose inconsistent means and ends, and that the feasibility of suggested strategies rarely has empirical backing. We argue that this vagueness actively hinders the technology from reaching its goals, as it encourages overclaiming, limits criticism, and prevents stakeholder feedback. Accordingly, we provide several recommendations for thinking and writing about the use of fact-checking artefacts. | [
"Schlichtkrull, Michael",
"Ousidhoum, Nedjma",
"Vlachos, Andreas"
] | The Intended Uses of Automated Fact-Checking Artefacts: Why, How and Who | findings-emnlp.577 | 2304.14238 | [
"https://github.com/michschli/intendedafcuses"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.578.bib | https://aclanthology.org/2023.findings-emnlp.578/ | @inproceedings{liu-etal-2023-retrieval,
title = "Retrieval-based Knowledge Transfer: An Effective Approach for Extreme Large Language Model Compression",
author = "Liu, Jiduan and
Liu, Jiahao and
Wang, Qifan and
Wang, Jingang and
Cai, Xunliang and
Zhao, Dongyan and
Wang, Ran and
Yan, Rui",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.578",
doi = "10.18653/v1/2023.findings-emnlp.578",
pages = "8643--8657",
abstract = "Large-scale pre-trained language models (LLMs) have demonstrated exceptional performance in various natural language processing (NLP) tasks. However, the massive size of these models poses huge challenges for their deployment in real-world applications. While numerous model compression techniques have been proposed, most of them are not well-suited for achieving extreme model compression when there is a significant gap in model scale. In this paper, we introduce a novel compression paradigm called Retrieval-based Knowledge Transfer (RetriKT), which effectively transfers the knowledge of LLMs to extremely small-scale models (e.g., 1{\%}). In particular, our approach extracts knowledge from LLMs to construct a knowledge store, from which the small-scale model can retrieve relevant information and leverage it for effective inference. To improve the quality of the model, soft prompt tuning and Proximal Policy Optimization (PPO) reinforcement learning techniques are employed. Extensive experiments are conducted on low-resource tasks from SuperGLUE and GLUE benchmarks. The results demonstrate that the proposed approach significantly enhances the performance of small-scale models by leveraging the knowledge from LLMs.",
}
| Large-scale pre-trained language models (LLMs) have demonstrated exceptional performance in various natural language processing (NLP) tasks. However, the massive size of these models poses huge challenges for their deployment in real-world applications. While numerous model compression techniques have been proposed, most of them are not well-suited for achieving extreme model compression when there is a significant gap in model scale. In this paper, we introduce a novel compression paradigm called Retrieval-based Knowledge Transfer (RetriKT), which effectively transfers the knowledge of LLMs to extremely small-scale models (e.g., 1{\%}). In particular, our approach extracts knowledge from LLMs to construct a knowledge store, from which the small-scale model can retrieve relevant information and leverage it for effective inference. To improve the quality of the model, soft prompt tuning and Proximal Policy Optimization (PPO) reinforcement learning techniques are employed. Extensive experiments are conducted on low-resource tasks from SuperGLUE and GLUE benchmarks. The results demonstrate that the proposed approach significantly enhances the performance of small-scale models by leveraging the knowledge from LLMs. | [
"Liu, Jiduan",
"Liu, Jiahao",
"Wang, Qifan",
"Wang, Jingang",
"Cai, Xunliang",
"Zhao, Dongyan",
"Wang, Ran",
"Yan, Rui"
] | Retrieval-based Knowledge Transfer: An Effective Approach for Extreme Large Language Model Compression | findings-emnlp.578 | 2310.15594 | [
""
] | https://huggingface.co/papers/2310.15594 | 0 | 1 | 0 | 8 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.579.bib | https://aclanthology.org/2023.findings-emnlp.579/ | @inproceedings{pour-etal-2023-count,
title = "{COUNT}: {CO}ntrastive {UN}likelihood Text Style Transfer for Text Detoxification",
author = "Pour, Mohammad Mahdi Abdollah and
Farinneya, Parsa and
Bharadwaj, Manasa and
Verma, Nikhil and
Pesaranghader, Ali and
Sanner, Scott",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.579",
doi = "10.18653/v1/2023.findings-emnlp.579",
pages = "8658--8666",
abstract = "Offensive and toxic text on social media platforms can lead to polarization and divisiveness within online communities and hinders constructive dialogue. Text detoxification is a crucial task in natural language processing to ensure the generation of non-toxic and safe text. Text detoxification is a special case of the Text Style Transfer (TST) problem, where an input text is rephrased to an output text that preserves its content while modifying the style (in this case to a more neutral, non-toxic style). State-of-the-art methods for detoxification use supervised training of encoder-decoder models to produce gold-standard outputs with a standard likelihood-based objective. However, it can be hard for these models to deviate from their pretrained auto-encoder identity mapping. While previous methods have used unlikelihood-based losses to penalize input-to-output copying of toxic content, these methods also unfortunately penalize non-toxic content in the input that would be fine to preserve in the output. To address these issues, we introduce a novel contrastive unlikelihood objective (COUNT) that directly contrasts the gold standard rephrasing with the identity input-to-output mapping to effectively isolate and focus learning on non-toxic style transfer. We benchmark COUNT on two parallel datasets, ParaDetox and APPDIA, showing that it achieves significant improvements in jointly combined fluency, content preservation, and detoxification (i.e., the highest {``}J{''} score).",
}
| Offensive and toxic text on social media platforms can lead to polarization and divisiveness within online communities and hinders constructive dialogue. Text detoxification is a crucial task in natural language processing to ensure the generation of non-toxic and safe text. Text detoxification is a special case of the Text Style Transfer (TST) problem, where an input text is rephrased to an output text that preserves its content while modifying the style (in this case to a more neutral, non-toxic style). State-of-the-art methods for detoxification use supervised training of encoder-decoder models to produce gold-standard outputs with a standard likelihood-based objective. However, it can be hard for these models to deviate from their pretrained auto-encoder identity mapping. While previous methods have used unlikelihood-based losses to penalize input-to-output copying of toxic content, these methods also unfortunately penalize non-toxic content in the input that would be fine to preserve in the output. To address these issues, we introduce a novel contrastive unlikelihood objective (COUNT) that directly contrasts the gold standard rephrasing with the identity input-to-output mapping to effectively isolate and focus learning on non-toxic style transfer. We benchmark COUNT on two parallel datasets, ParaDetox and APPDIA, showing that it achieves significant improvements in jointly combined fluency, content preservation, and detoxification (i.e., the highest {``}J{''} score). | [
"Pour, Mohammad Mahdi Abdollah",
"Farinneya, Parsa",
"Bharadwaj, Manasa",
"Verma, Nikhil",
"Pesaranghader, Ali",
"Sanner, Scott"
] | COUNT: COntrastive UNlikelihood Text Style Transfer for Text Detoxification | findings-emnlp.579 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.580.bib | https://aclanthology.org/2023.findings-emnlp.580/ | @inproceedings{wei-etal-2023-kicgpt,
title = "{KICGPT}: Large Language Model with Knowledge in Context for Knowledge Graph Completion",
author = "Wei, Yanbin and
Huang, Qiushi and
Zhang, Yu and
Kwok, James",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.580",
doi = "10.18653/v1/2023.findings-emnlp.580",
pages = "8667--8683",
abstract = "Knowledge Graph Completion (KGC) is crucial for addressing knowledge graph incompleteness and supporting downstream applications. Many models have been proposed for KGC and they can be categorized into two main classes, including triple-based and test-based approaches. Triple-based methods struggle with long-tail entities due to limited structural information and imbalanced distributions of entities. Text-based methods alleviate this issue but require costly training for language models and specific finetuning for knowledge graphs, which limits their efficiency. To alleviate the limitations in the two approaches, in this paper, we propose KICGPT, a framework that integrates a large language model (LLM) and a triple-based KGC retriever, to alleviate the long-tail problem without incurring additional training overhead. In the proposed KICGPT model, we propose an in-context learning strategy called Knowledge Prompt, which encodes structural knowledge into demonstrations to guide LLM. Empirical results on benchmark datasets demonstrate the effectiveness of the proposed KICGPT model with lighter training overhead and no finetuning.",
}
| Knowledge Graph Completion (KGC) is crucial for addressing knowledge graph incompleteness and supporting downstream applications. Many models have been proposed for KGC and they can be categorized into two main classes, including triple-based and test-based approaches. Triple-based methods struggle with long-tail entities due to limited structural information and imbalanced distributions of entities. Text-based methods alleviate this issue but require costly training for language models and specific finetuning for knowledge graphs, which limits their efficiency. To alleviate the limitations in the two approaches, in this paper, we propose KICGPT, a framework that integrates a large language model (LLM) and a triple-based KGC retriever, to alleviate the long-tail problem without incurring additional training overhead. In the proposed KICGPT model, we propose an in-context learning strategy called Knowledge Prompt, which encodes structural knowledge into demonstrations to guide LLM. Empirical results on benchmark datasets demonstrate the effectiveness of the proposed KICGPT model with lighter training overhead and no finetuning. | [
"Wei, Yanbin",
"Huang, Qiushi",
"Zhang, Yu",
"Kwok, James"
] | KICGPT: Large Language Model with Knowledge in Context for Knowledge Graph Completion | findings-emnlp.580 | 2402.02389 | [
"https://github.com/weiyanbin1999/kicgpt"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.581.bib | https://aclanthology.org/2023.findings-emnlp.581/ | @inproceedings{zhang-etal-2023-show,
title = "Show, Write, and Retrieve: Entity-aware Article Generation and Retrieval",
author = "Zhang, Zhongping and
Gu, Yiwen and
Plummer, Bryan",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.581",
doi = "10.18653/v1/2023.findings-emnlp.581",
pages = "8684--8704",
abstract = "Article comprehension is an important challenge in natural language processing with many applications such as article generation or image-to-article retrieval. Prior work typically encodes all tokens in articles uniformly using pretrained language models. However, in many applications, such as understanding news stories, these articles are based on real-world events and may reference many named entities that are difficult to accurately recognize and predict by language models. To address this challenge, we propose an ENtity-aware article GeneratIoN and rEtrieval (ENGINE) framework, to explicitly incorporate named entities into language models. ENGINE has two main components: a named-entity extraction module to extract named entities from both metadata and embedded images associated with articles, and an entity-aware mechanism that enhances the model{'}s ability to recognize and predict entity names. We conducted experiments on three public datasets: GoodNews, VisualNews, and WikiText, where our results demonstrate that our model can boost both article generation and article retrieval performance, with a 4-5 perplexity improvement in article generation and a 3-4{\%} boost in recall@1 in article retrieval. We release our implementation at [this http URL](https://github.com/Zhongping-Zhang/ENGINE).",
}
| Article comprehension is an important challenge in natural language processing with many applications such as article generation or image-to-article retrieval. Prior work typically encodes all tokens in articles uniformly using pretrained language models. However, in many applications, such as understanding news stories, these articles are based on real-world events and may reference many named entities that are difficult to accurately recognize and predict by language models. To address this challenge, we propose an ENtity-aware article GeneratIoN and rEtrieval (ENGINE) framework, to explicitly incorporate named entities into language models. ENGINE has two main components: a named-entity extraction module to extract named entities from both metadata and embedded images associated with articles, and an entity-aware mechanism that enhances the model{'}s ability to recognize and predict entity names. We conducted experiments on three public datasets: GoodNews, VisualNews, and WikiText, where our results demonstrate that our model can boost both article generation and article retrieval performance, with a 4-5 perplexity improvement in article generation and a 3-4{\%} boost in recall@1 in article retrieval. We release our implementation at [this http URL](https://github.com/Zhongping-Zhang/ENGINE). | [
"Zhang, Zhongping",
"Gu, Yiwen",
"Plummer, Bryan"
] | Show, Write, and Retrieve: Entity-aware Article Generation and Retrieval | findings-emnlp.581 | 2112.05917 | [
"https://github.com/zhongping-zhang/engine"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.582.bib | https://aclanthology.org/2023.findings-emnlp.582/ | @inproceedings{timkey-linzen-2023-language,
title = "A Language Model with Limited Memory Capacity Captures Interference in Human Sentence Processing",
author = "Timkey, William and
Linzen, Tal",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.582",
doi = "10.18653/v1/2023.findings-emnlp.582",
pages = "8705--8720",
abstract = "Two of the central factors believed to underpin human sentence processing difficulty are expectations and retrieval from working memory. A recent attempt to create a unified cognitive model integrating these two factors have relied on the parallels between the self-attention mechanism of transformer language models and cue-based retrieval theories of working memory in human sentence processing (Ryu and Lewis 2021). While the authors show that attention patterns in specialized attention heads of GPT-2 are consistent with a key prediction of cue-based retrieval models, similarity-based interference effects, their method requires the identification of syntactically specialized attention heads, and makes an cognitively implausible implicit assumption that hundreds of memory retrieval operations take place in parallel. In the present work, we develop a recurrent neural language model with a single self-attention head, which more closely parallels the memory system assumed by cognitive theories. We show that our model{'}s single attention head can capture semantic and syntactic interference effects observed in human experiments.",
}
| Two of the central factors believed to underpin human sentence processing difficulty are expectations and retrieval from working memory. A recent attempt to create a unified cognitive model integrating these two factors have relied on the parallels between the self-attention mechanism of transformer language models and cue-based retrieval theories of working memory in human sentence processing (Ryu and Lewis 2021). While the authors show that attention patterns in specialized attention heads of GPT-2 are consistent with a key prediction of cue-based retrieval models, similarity-based interference effects, their method requires the identification of syntactically specialized attention heads, and makes an cognitively implausible implicit assumption that hundreds of memory retrieval operations take place in parallel. In the present work, we develop a recurrent neural language model with a single self-attention head, which more closely parallels the memory system assumed by cognitive theories. We show that our model{'}s single attention head can capture semantic and syntactic interference effects observed in human experiments. | [
"Timkey, William",
"Linzen, Tal"
] | A Language Model with Limited Memory Capacity Captures Interference in Human Sentence Processing | findings-emnlp.582 | 2310.16142 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.583.bib | https://aclanthology.org/2023.findings-emnlp.583/ | @inproceedings{fang-etal-2023-annotations,
title = "Annotations Are Not All You Need: A Cross-modal Knowledge Transfer Network for Unsupervised Temporal Sentence Grounding",
author = "Fang, Xiang and
Liu, Daizong and
Fang, Wanlong and
Zhou, Pan and
Cheng, Yu and
Tang, Keke and
Zou, Kai",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.583",
doi = "10.18653/v1/2023.findings-emnlp.583",
pages = "8721--8733",
abstract = "This paper addresses the task of temporal sentence grounding (TSG). Although many respectable works have made decent achievements in this important topic, they severely rely on massive expensive video-query paired annotations, which require a tremendous amount of human effort to collect in real-world applications. To this end, in this paper, we target a more practical but challenging TSG setting: unsupervised temporal sentence grounding, where both paired video-query and segment boundary annotations are unavailable during the network training. Considering that some other cross-modal tasks provide many easily available yet cheap labels, we tend to collect and transfer their simple cross-modal alignment knowledge into our complex scenarios: 1) We first explore the entity-aware object-guided appearance knowledge from the paired Image-Noun task, and adapt them into each independent video frame; 2) Then, we extract the event-aware action representation from the paired Video-Verb task, and further refine the action representation into more practical but complicated real-world cases by a newly proposed copy-paste approach; 3) By modulating and transferring both appearance and action knowledge into our challenging unsupervised task, our model can directly utilize this general knowledge to correlate videos and queries, and accurately retrieve the relevant segment without training. Extensive experiments on two challenging datasets (ActivityNet Captions and Charades-STA) show our effectiveness, outperforming existing unsupervised methods and even competitively beating supervised works.",
}
| This paper addresses the task of temporal sentence grounding (TSG). Although many respectable works have made decent achievements in this important topic, they severely rely on massive expensive video-query paired annotations, which require a tremendous amount of human effort to collect in real-world applications. To this end, in this paper, we target a more practical but challenging TSG setting: unsupervised temporal sentence grounding, where both paired video-query and segment boundary annotations are unavailable during the network training. Considering that some other cross-modal tasks provide many easily available yet cheap labels, we tend to collect and transfer their simple cross-modal alignment knowledge into our complex scenarios: 1) We first explore the entity-aware object-guided appearance knowledge from the paired Image-Noun task, and adapt them into each independent video frame; 2) Then, we extract the event-aware action representation from the paired Video-Verb task, and further refine the action representation into more practical but complicated real-world cases by a newly proposed copy-paste approach; 3) By modulating and transferring both appearance and action knowledge into our challenging unsupervised task, our model can directly utilize this general knowledge to correlate videos and queries, and accurately retrieve the relevant segment without training. Extensive experiments on two challenging datasets (ActivityNet Captions and Charades-STA) show our effectiveness, outperforming existing unsupervised methods and even competitively beating supervised works. | [
"Fang, Xiang",
"Liu, Daizong",
"Fang, Wanlong",
"Zhou, Pan",
"Cheng, Yu",
"Tang, Keke",
"Zou, Kai"
] | Annotations Are Not All You Need: A Cross-modal Knowledge Transfer Network for Unsupervised Temporal Sentence Grounding | findings-emnlp.583 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.584.bib | https://aclanthology.org/2023.findings-emnlp.584/ | @inproceedings{wu-etal-2023-parameter,
title = "Parameter Efficient Multi-task Fine-tuning by Learning to Transfer Token-wise Prompts",
author = "Wu, Muling and
Liu, Wenhao and
Xu, Jianhan and
Lv, Changze and
Ling, Zixuan and
Li, Tianlong and
Huang, Longtao and
Zheng, Xiaoqing and
Huang, Xuanjing",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.584",
doi = "10.18653/v1/2023.findings-emnlp.584",
pages = "8734--8746",
abstract = "Prompt tuning has been proven to be successful on various tasks by incorporating a small number of trainable parameters while freezing large pre-trained language models (PLMs). However, it is still unsettled how to generate more proper prompts for any individual examples and how to extend prompt tuning to multi-task learning scenarios by leveraging cross-task features. To address these challenges, we propose a token-wise prompt tuning (TPT), in which a bank of finer-grained soft prompt tokens is built for multi-task learning by memory network. The tokens are retrieved from the bank against an input example and assembled to an instance-dependent prompt. Extensive experimental results on 14 datasets demonstrated that the models enhanced by our TPT performed far better than full parameter fine-tuned models and achieved state-of-the-art by tuning only 0.035{\%} parameters.",
}
| Prompt tuning has been proven to be successful on various tasks by incorporating a small number of trainable parameters while freezing large pre-trained language models (PLMs). However, it is still unsettled how to generate more proper prompts for any individual examples and how to extend prompt tuning to multi-task learning scenarios by leveraging cross-task features. To address these challenges, we propose a token-wise prompt tuning (TPT), in which a bank of finer-grained soft prompt tokens is built for multi-task learning by memory network. The tokens are retrieved from the bank against an input example and assembled to an instance-dependent prompt. Extensive experimental results on 14 datasets demonstrated that the models enhanced by our TPT performed far better than full parameter fine-tuned models and achieved state-of-the-art by tuning only 0.035{\%} parameters. | [
"Wu, Muling",
"Liu, Wenhao",
"Xu, Jianhan",
"Lv, Changze",
"Ling, Zixuan",
"Li, Tianlong",
"Huang, Longtao",
"Zheng, Xiaoqing",
"Huang, Xuanjing"
] | Parameter Efficient Multi-task Fine-tuning by Learning to Transfer Token-wise Prompts | findings-emnlp.584 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.585.bib | https://aclanthology.org/2023.findings-emnlp.585/ | @inproceedings{veloso-etal-2023-rewriting,
title = "A Rewriting Approach for Gender Inclusivity in {P}ortuguese",
author = "Veloso, Leonor and
Coheur, Luisa and
Ribeiro, Rui",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.585",
doi = "10.18653/v1/2023.findings-emnlp.585",
pages = "8747--8759",
abstract = "In recent years, there has been a notable rise in research interest regarding the integration of gender-inclusive and gender-neutral language in natural language processing models. A specific area of focus that has gained practical and academic significant interest is gender-neutral rewriting, which involves converting binary-gendered text to its gender-neutral counterpart. However, current approaches to gender-neutral rewriting for gendered languages tend to rely on large datasets, which may not be an option for languages with fewer resources, such as Portuguese. In this paper, we present a rule-based and a neural-based tool for gender-neutral rewriting for Portuguese, a heavily gendered Romance language whose morphology creates different challenges from the ones tackled by other gender-neutral rewriters. Our neural approach relies on fine-tuning large multilingual machine translation models on examples generated by the rule-based model. We evaluate both models on texts from different sources and contexts. We provide the first Portuguese dataset explicitly containing gender-neutral language and neopronouns, as well as a manually annotated golden collection of 500 sentences that allows for evaluation of future work.",
}
| In recent years, there has been a notable rise in research interest regarding the integration of gender-inclusive and gender-neutral language in natural language processing models. A specific area of focus that has gained practical and academic significant interest is gender-neutral rewriting, which involves converting binary-gendered text to its gender-neutral counterpart. However, current approaches to gender-neutral rewriting for gendered languages tend to rely on large datasets, which may not be an option for languages with fewer resources, such as Portuguese. In this paper, we present a rule-based and a neural-based tool for gender-neutral rewriting for Portuguese, a heavily gendered Romance language whose morphology creates different challenges from the ones tackled by other gender-neutral rewriters. Our neural approach relies on fine-tuning large multilingual machine translation models on examples generated by the rule-based model. We evaluate both models on texts from different sources and contexts. We provide the first Portuguese dataset explicitly containing gender-neutral language and neopronouns, as well as a manually annotated golden collection of 500 sentences that allows for evaluation of future work. | [
"Veloso, Leonor",
"Coheur, Luisa",
"Ribeiro, Rui"
] | A Rewriting Approach for Gender Inclusivity in Portuguese | findings-emnlp.585 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.586.bib | https://aclanthology.org/2023.findings-emnlp.586/ | @inproceedings{xiong-etal-2023-eara,
title = "{EARA}: Improving Biomedical Semantic Textual Similarity with Entity-Aligned Attention and Retrieval Augmentation",
author = "Xiong, Ying and
Yang, Xin and
Liu, Linjing and
Wong, Ka-Chun and
Chen, Qingcai and
Xiang, Yang and
Tang, Buzhou",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.586",
doi = "10.18653/v1/2023.findings-emnlp.586",
pages = "8760--8771",
abstract = "Measuring Semantic Textual Similarity (STS) is a fundamental task in biomedical text processing, which aims at quantifying the similarity between two input biomedical sentences. Unfortunately, the STS datasets in the biomedical domain are relatively smaller but more complex in semantics than common domain, often leading to overfitting issues and insufficient text representation even based on Pre-trained Language Models (PLMs) due to too many biomedical entities. In this paper, we propose EARA, an entity-aligned, attention-based and retrieval-augmented PLMs. Our proposed EARA first aligns the same type of fine-grained entity information in each sentence pair with an entity alignment matrix. Then, EARA regularizes the attention mechanism with an entity alignment matrix with an auxiliary loss. Finally, we add a retrieval module that retrieves similar instances to expand the scope of entity pairs and improve the model{'}s generalization. The comprehensive experiments reflect that EARA can achieve state-of-the-art performance on both in-domain and out-of-domain datasets. Source code is available.",
}
| Measuring Semantic Textual Similarity (STS) is a fundamental task in biomedical text processing, which aims at quantifying the similarity between two input biomedical sentences. Unfortunately, the STS datasets in the biomedical domain are relatively smaller but more complex in semantics than common domain, often leading to overfitting issues and insufficient text representation even based on Pre-trained Language Models (PLMs) due to too many biomedical entities. In this paper, we propose EARA, an entity-aligned, attention-based and retrieval-augmented PLMs. Our proposed EARA first aligns the same type of fine-grained entity information in each sentence pair with an entity alignment matrix. Then, EARA regularizes the attention mechanism with an entity alignment matrix with an auxiliary loss. Finally, we add a retrieval module that retrieves similar instances to expand the scope of entity pairs and improve the model{'}s generalization. The comprehensive experiments reflect that EARA can achieve state-of-the-art performance on both in-domain and out-of-domain datasets. Source code is available. | [
"Xiong, Ying",
"Yang, Xin",
"Liu, Linjing",
"Wong, Ka-Chun",
"Chen, Qingcai",
"Xiang, Yang",
"Tang, Buzhou"
] | EARA: Improving Biomedical Semantic Textual Similarity with Entity-Aligned Attention and Retrieval Augmentation | findings-emnlp.586 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.587.bib | https://aclanthology.org/2023.findings-emnlp.587/ | @inproceedings{zhang-etal-2023-neuro,
title = "Neuro-Symbolic Sentiment Analysis with Dynamic Word Sense Disambiguation",
author = "Zhang, Xulang and
Mao, Rui and
He, Kai and
Cambria, Erik",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.587",
doi = "10.18653/v1/2023.findings-emnlp.587",
pages = "8772--8783",
abstract = "Sentiment analysis is a task that highly depends on the understanding of word senses. Traditional neural network models are black boxes that represent word senses as vectors that are uninterpretable for humans. On the other hand, the application of Word Sense Disambiguation (WSD) systems in downstream tasks poses challenges regarding i) which words need to be disambiguated, and ii) how to model explicit word senses into easily understandable terms for a downstream model. This work proposes a neurosymbolic framework that incorporates WSD by identifying and paraphrasing ambiguous words to improve the accuracy of sentiment predictions. The framework allows us to understand which words are paraphrased into which semantically unequivocal words, thus enabling a downstream task model to gain both accuracy and interpretability. To better fine-tune a lexical substitution model for WSD on a downstream task without ground-truth word sense labels, we leverage dynamic rewarding to jointly train sentiment analysis and lexical substitution models. Our framework proves to effectively improve the performance of sentiment analysis on corpora from different domains.",
}
| Sentiment analysis is a task that highly depends on the understanding of word senses. Traditional neural network models are black boxes that represent word senses as vectors that are uninterpretable for humans. On the other hand, the application of Word Sense Disambiguation (WSD) systems in downstream tasks poses challenges regarding i) which words need to be disambiguated, and ii) how to model explicit word senses into easily understandable terms for a downstream model. This work proposes a neurosymbolic framework that incorporates WSD by identifying and paraphrasing ambiguous words to improve the accuracy of sentiment predictions. The framework allows us to understand which words are paraphrased into which semantically unequivocal words, thus enabling a downstream task model to gain both accuracy and interpretability. To better fine-tune a lexical substitution model for WSD on a downstream task without ground-truth word sense labels, we leverage dynamic rewarding to jointly train sentiment analysis and lexical substitution models. Our framework proves to effectively improve the performance of sentiment analysis on corpora from different domains. | [
"Zhang, Xulang",
"Mao, Rui",
"He, Kai",
"Cambria, Erik"
] | Neuro-Symbolic Sentiment Analysis with Dynamic Word Sense Disambiguation | findings-emnlp.587 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.588.bib | https://aclanthology.org/2023.findings-emnlp.588/ | @inproceedings{hronsky-keuleers-2023-role,
title = "Role of Context in Unsupervised Sentence Representation Learning: the Case of Dialog Act Modeling",
author = "Hronsky, Rastislav and
Keuleers, Emmanuel",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.588",
doi = "10.18653/v1/2023.findings-emnlp.588",
pages = "8784--8792",
abstract = "Unsupervised learning of word representations involves capturing the contextual information surrounding word occurrences, which can be grounded in the observation that word form is largely disconnected from word meaning. While there are fewer reasons to believe that the same holds for sentences, learning through context has been carried over to learning representations of word sequences. However, this work pays minimal to no attention to the role of context in inferring sentence representations. In this article, we present a dialog act tag probing task designed to explicitly compare content-, and context-oriented sentence representations inferred on utterances of telephone conversations (SwDA). Our results suggest that there is no clear benefit of context-based sentence representations over content-based sentence representations. However, there is a very clear benefit of increasing the dimensionality of the sentence vectors in nearly all approaches.",
}
| Unsupervised learning of word representations involves capturing the contextual information surrounding word occurrences, which can be grounded in the observation that word form is largely disconnected from word meaning. While there are fewer reasons to believe that the same holds for sentences, learning through context has been carried over to learning representations of word sequences. However, this work pays minimal to no attention to the role of context in inferring sentence representations. In this article, we present a dialog act tag probing task designed to explicitly compare content-, and context-oriented sentence representations inferred on utterances of telephone conversations (SwDA). Our results suggest that there is no clear benefit of context-based sentence representations over content-based sentence representations. However, there is a very clear benefit of increasing the dimensionality of the sentence vectors in nearly all approaches. | [
"Hronsky, Rastislav",
"Keuleers, Emmanuel"
] | Role of Context in Unsupervised Sentence Representation Learning: the Case of Dialog Act Modeling | findings-emnlp.588 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.589.bib | https://aclanthology.org/2023.findings-emnlp.589/ | @inproceedings{nandy-etal-2023-clmsm,
title = "{CLMSM}: A Multi-Task Learning Framework for Pre-training on Procedural Text",
author = "Nandy, Abhilash and
Kapadnis, Manav and
Goyal, Pawan and
Ganguly, Niloy",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.589",
doi = "10.18653/v1/2023.findings-emnlp.589",
pages = "8793--8806",
abstract = "In this paper, we propose ***CLMSM***, a domain-specific, continual pre-training framework, that learns from a large set of procedural recipes. ***CLMSM*** uses a Multi-Task Learning Framework to optimize two objectives - a) Contrastive Learning using hard triplets to learn fine-grained differences across entities in the procedures, and b) a novel Mask-Step Modelling objective to learn step-wise context of a procedure. We test the performance of ***CLMSM*** on the downstream tasks of tracking entities and aligning actions between two procedures on three datasets, one of which is an open-domain dataset not conforming with the pre-training dataset. We show that ***CLMSM*** not only outperforms baselines on recipes (in-domain) but is also able to generalize to open-domain procedural NLP tasks.",
}
| In this paper, we propose ***CLMSM***, a domain-specific, continual pre-training framework, that learns from a large set of procedural recipes. ***CLMSM*** uses a Multi-Task Learning Framework to optimize two objectives - a) Contrastive Learning using hard triplets to learn fine-grained differences across entities in the procedures, and b) a novel Mask-Step Modelling objective to learn step-wise context of a procedure. We test the performance of ***CLMSM*** on the downstream tasks of tracking entities and aligning actions between two procedures on three datasets, one of which is an open-domain dataset not conforming with the pre-training dataset. We show that ***CLMSM*** not only outperforms baselines on recipes (in-domain) but is also able to generalize to open-domain procedural NLP tasks. | [
"N",
"y, Abhilash",
"Kapadnis, Manav",
"Goyal, Pawan",
"Ganguly, Niloy"
] | CLMSM: A Multi-Task Learning Framework for Pre-training on Procedural Text | findings-emnlp.589 | 2310.14326 | [
"https://github.com/manavkapadnis/clmsm_emnlp_2023"
] | https://huggingface.co/papers/2310.14326 | 1 | 1 | 0 | 4 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.590.bib | https://aclanthology.org/2023.findings-emnlp.590/ | @inproceedings{zhuang-etal-2023-open,
title = "Open-source Large Language Models are Strong Zero-shot Query Likelihood Models for Document Ranking",
author = "Zhuang, Shengyao and
Liu, Bing and
Koopman, Bevan and
Zuccon, Guido",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.590",
doi = "10.18653/v1/2023.findings-emnlp.590",
pages = "8807--8817",
abstract = "In the field of information retrieval, Query Likelihood Models (QLMs) rank documents based on the probability of generating the query given the content of a document. Recently, advanced large language models (LLMs) have emerged as effective QLMs, showcasing promising ranking capabilities. This paper focuses on investigating the genuine zero-shot ranking effectiveness of recent LLMs, which are solely pre-trained on unstructured text data without supervised instruction fine-tuning. Our findings reveal the robust zero-shot ranking ability of such LLMs, highlighting that additional instruction fine-tuning may hinder effectiveness unless a question generation task is present in the fine-tuning dataset. Furthermore, we introduce a novel state-of-the-art ranking system that integrates LLM-based QLMs with a hybrid zero-shot retriever, demonstrating exceptional effectiveness in both zero-shot and few-shot scenarios. We make our codebase publicly available at https://github.com/ielab/llm-qlm.",
}
| In the field of information retrieval, Query Likelihood Models (QLMs) rank documents based on the probability of generating the query given the content of a document. Recently, advanced large language models (LLMs) have emerged as effective QLMs, showcasing promising ranking capabilities. This paper focuses on investigating the genuine zero-shot ranking effectiveness of recent LLMs, which are solely pre-trained on unstructured text data without supervised instruction fine-tuning. Our findings reveal the robust zero-shot ranking ability of such LLMs, highlighting that additional instruction fine-tuning may hinder effectiveness unless a question generation task is present in the fine-tuning dataset. Furthermore, we introduce a novel state-of-the-art ranking system that integrates LLM-based QLMs with a hybrid zero-shot retriever, demonstrating exceptional effectiveness in both zero-shot and few-shot scenarios. We make our codebase publicly available at https://github.com/ielab/llm-qlm. | [
"Zhuang, Shengyao",
"Liu, Bing",
"Koopman, Bevan",
"Zuccon, Guido"
] | Open-source Large Language Models are Strong Zero-shot Query Likelihood Models for Document Ranking | findings-emnlp.590 | 2310.13243 | [
"https://github.com/ielab/llm-qlm"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.591.bib | https://aclanthology.org/2023.findings-emnlp.591/ | @inproceedings{schlangen-2023-general,
title = "On General Language Understanding",
author = "Schlangen, David",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.591",
doi = "10.18653/v1/2023.findings-emnlp.591",
pages = "8818--8825",
abstract = "Natural Language Processing prides itself to be an empirically-minded, if not outright empiricist field, and yet lately it seems to get itself into essentialist debates on issues of meaning and measurement ({``}Do Large Language Models Understand Language, And If So, How Much?{''}). This is not by accident: Here, as everywhere, the evidence underspecifies the understanding. As a remedy, this paper sketches the outlines of a model of understanding, which can ground questions of the adequacy of current methods of measurement of model quality. The paper makes three claims: A) That different language use situation types have different characteristics, B) That language understanding is a multifaceted phenomenon, bringing together individualistic and social processes, and C) That the choice of Understanding Indicator marks the limits of benchmarking, and the beginnings of considerations of the ethics of NLP use.",
}
| Natural Language Processing prides itself to be an empirically-minded, if not outright empiricist field, and yet lately it seems to get itself into essentialist debates on issues of meaning and measurement ({``}Do Large Language Models Understand Language, And If So, How Much?{''}). This is not by accident: Here, as everywhere, the evidence underspecifies the understanding. As a remedy, this paper sketches the outlines of a model of understanding, which can ground questions of the adequacy of current methods of measurement of model quality. The paper makes three claims: A) That different language use situation types have different characteristics, B) That language understanding is a multifaceted phenomenon, bringing together individualistic and social processes, and C) That the choice of Understanding Indicator marks the limits of benchmarking, and the beginnings of considerations of the ethics of NLP use. | [
"Schlangen, David"
] | On General Language Understanding | findings-emnlp.591 | 2310.18038 | [
""
] | https://huggingface.co/papers/2407.10972 | 1 | 0 | 0 | 4 | [] | [
"vgbench/VGQA"
] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.592.bib | https://aclanthology.org/2023.findings-emnlp.592/ | @inproceedings{krishna-etal-2023-usb,
title = "{USB}: A Unified Summarization Benchmark Across Tasks and Domains",
author = "Krishna, Kundan and
Gupta, Prakhar and
Ramprasad, Sanjana and
Wallace, Byron and
Bigham, Jeffrey and
Lipton, Zachary",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.592",
doi = "10.18653/v1/2023.findings-emnlp.592",
pages = "8826--8845",
abstract = "While the NLP community has produced numerous summarization benchmarks, none provide the rich annotations required to simultaneously address many important problems related to control and reliability. We introduce a Wikipedia-derived benchmark, complemented by a rich set of crowd-sourced annotations, that supports 8 interrelated tasks: (i) extractive summarization; (ii) abstractive summarization; (iii) topic-based summarization; (iv) compressing selected sentences into a one-line summary; (v) surfacing evidence for a summary sentence; (vi) predicting the factual accuracy of a summary sentence; (vii) identifying unsubstantiated spans in a summary sentence; (viii) correcting factual errors in summaries. We compare various methods on this benchmark and discover that on multiple tasks, moderately-sized fine-tuned models consistently outperform much larger few-shot prompted language models. For factuality-related tasks, we also evaluate existing heuristics to create training data and find that training on them results in worse performance than training on $20\times$ less human-labeled data. Our articles draw from 6 domains, facilitating cross-domain analysis. On some tasks, the amount of training data matters more than the domain where it comes from, while for other tasks training specifically on data from the target domain, even if limited, is more beneficial.",
}
| While the NLP community has produced numerous summarization benchmarks, none provide the rich annotations required to simultaneously address many important problems related to control and reliability. We introduce a Wikipedia-derived benchmark, complemented by a rich set of crowd-sourced annotations, that supports 8 interrelated tasks: (i) extractive summarization; (ii) abstractive summarization; (iii) topic-based summarization; (iv) compressing selected sentences into a one-line summary; (v) surfacing evidence for a summary sentence; (vi) predicting the factual accuracy of a summary sentence; (vii) identifying unsubstantiated spans in a summary sentence; (viii) correcting factual errors in summaries. We compare various methods on this benchmark and discover that on multiple tasks, moderately-sized fine-tuned models consistently outperform much larger few-shot prompted language models. For factuality-related tasks, we also evaluate existing heuristics to create training data and find that training on them results in worse performance than training on $20\times$ less human-labeled data. Our articles draw from 6 domains, facilitating cross-domain analysis. On some tasks, the amount of training data matters more than the domain where it comes from, while for other tasks training specifically on data from the target domain, even if limited, is more beneficial. | [
"Krishna, Kundan",
"Gupta, Prakhar",
"Ramprasad, Sanjana",
"Wallace, Byron",
"Bigham, Jeffrey",
"Lipton, Zachary"
] | USB: A Unified Summarization Benchmark Across Tasks and Domains | findings-emnlp.592 | 2305.14296 | [
"https://github.com/kukrishna/usb"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.593.bib | https://aclanthology.org/2023.findings-emnlp.593/ | @inproceedings{sarkar-etal-2023-tage,
title = "tag{E}: Enabling an Embodied Agent to Understand Human Instructions",
author = "Sarkar, Chayan and
Mitra, Avik and
Pramanick, Pradip and
Nayak, Tapas",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.593",
doi = "10.18653/v1/2023.findings-emnlp.593",
pages = "8846--8857",
abstract = "Natural language serves as the primary mode of communication when an intelligent agent with a physical presence engages with human beings. While a plethora of research focuses on natural language understanding (NLU), encompassing endeavors such as sentiment analysis, intent prediction, question answering, and summarization, the scope of NLU directed at situations necessitating tangible actions by an embodied agent remains limited. The inherent ambiguity and incompleteness inherent in natural language present challenges for intelligent agents striving to decipher human intention. To tackle this predicament head-on, we introduce a novel system known as task and argument grounding for Embodied agents (tagE). At its core, our system employs an inventive neural network model designed to extract a series of tasks from complex task instructions expressed in natural language. Our proposed model adopts an encoder-decoder framework enriched with nested decoding to effectively extract tasks and their corresponding arguments from these intricate instructions. These extracted tasks are then mapped (or grounded) to the robot{'}s established collection of skills, while the arguments find grounding in objects present within the environment. To facilitate the training and evaluation of our system, we have curated a dataset featuring complex instructions. The results of our experiments underscore the prowess of our approach, as it outperforms robust baseline models.",
}
| Natural language serves as the primary mode of communication when an intelligent agent with a physical presence engages with human beings. While a plethora of research focuses on natural language understanding (NLU), encompassing endeavors such as sentiment analysis, intent prediction, question answering, and summarization, the scope of NLU directed at situations necessitating tangible actions by an embodied agent remains limited. The inherent ambiguity and incompleteness inherent in natural language present challenges for intelligent agents striving to decipher human intention. To tackle this predicament head-on, we introduce a novel system known as task and argument grounding for Embodied agents (tagE). At its core, our system employs an inventive neural network model designed to extract a series of tasks from complex task instructions expressed in natural language. Our proposed model adopts an encoder-decoder framework enriched with nested decoding to effectively extract tasks and their corresponding arguments from these intricate instructions. These extracted tasks are then mapped (or grounded) to the robot{'}s established collection of skills, while the arguments find grounding in objects present within the environment. To facilitate the training and evaluation of our system, we have curated a dataset featuring complex instructions. The results of our experiments underscore the prowess of our approach, as it outperforms robust baseline models. | [
"Sarkar, Chayan",
"Mitra, Avik",
"Pramanick, Pradip",
"Nayak, Tapas"
] | tagE: Enabling an Embodied Agent to Understand Human Instructions | findings-emnlp.593 | 2310.15605 | [
"https://github.com/csarkar/tage"
] | https://huggingface.co/papers/2310.15605 | 1 | 0 | 0 | 4 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.594.bib | https://aclanthology.org/2023.findings-emnlp.594/ | @inproceedings{yu-etal-2023-instances,
title = "Instances and Labels: Hierarchy-aware Joint Supervised Contrastive Learning for Hierarchical Multi-Label Text Classification",
author = "Yu, Simon Chi Lok and
He, Jie and
Basulto, Victor and
Pan, Jeff",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.594",
doi = "10.18653/v1/2023.findings-emnlp.594",
pages = "8858--8875",
abstract = "Hierarchical multi-label text classification (HMTC) aims at utilizing a label hierarchy in multi-label classification. Recent approaches to HMTC deal with the problem of imposing an overconstrained premise on the output space by using contrastive learning on generated samples in a semi-supervised manner to bring text and label embeddings closer. However, the generation of samples tends to introduce noise as it ignores the correlation between similar samples in the same batch. One solution to this issue is supervised contrastive learning, but it remains an underexplored topic in HMTC due to its complex structured labels. To overcome this challenge, we propose **HJCL**, a **H**ierarchy-aware **J**oint Supervised **C**ontrastive **L**earning method that bridges the gap between supervised contrastive learning and HMTC. Specifically, we employ both instance-wise and label-wise contrastive learning techniques and carefully construct batches to fulfill the contrastive learning objective. Extensive experiments on four multi-path HMTC datasets demonstrate that HJCLachieves promising results and the effectiveness of Contrastive Learning on HMTC. Code and data are available at https://github.com/simonucl/HJCL.",
}
| Hierarchical multi-label text classification (HMTC) aims at utilizing a label hierarchy in multi-label classification. Recent approaches to HMTC deal with the problem of imposing an overconstrained premise on the output space by using contrastive learning on generated samples in a semi-supervised manner to bring text and label embeddings closer. However, the generation of samples tends to introduce noise as it ignores the correlation between similar samples in the same batch. One solution to this issue is supervised contrastive learning, but it remains an underexplored topic in HMTC due to its complex structured labels. To overcome this challenge, we propose **HJCL**, a **H**ierarchy-aware **J**oint Supervised **C**ontrastive **L**earning method that bridges the gap between supervised contrastive learning and HMTC. Specifically, we employ both instance-wise and label-wise contrastive learning techniques and carefully construct batches to fulfill the contrastive learning objective. Extensive experiments on four multi-path HMTC datasets demonstrate that HJCLachieves promising results and the effectiveness of Contrastive Learning on HMTC. Code and data are available at https://github.com/simonucl/HJCL. | [
"Yu, Simon Chi Lok",
"He, Jie",
"Basulto, Victor",
"Pan, Jeff"
] | Instances and Labels: Hierarchy-aware Joint Supervised Contrastive Learning for Hierarchical Multi-Label Text Classification | findings-emnlp.594 | 2310.05128 | [
"https://github.com/simonucl/HJCL"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.595.bib | https://aclanthology.org/2023.findings-emnlp.595/ | @inproceedings{feng-etal-2023-uncovering,
title = "Uncovering Limitations in Text-to-Image Generation: A Contrastive Approach with Structured Semantic Alignment",
author = "Feng, Qianyu and
Sui, Yulei and
Zhang, Hongyu",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.595",
doi = "10.18653/v1/2023.findings-emnlp.595",
pages = "8876--8888",
abstract = "Despite significant advancements in text-to-image generation models, they still face challenges when it comes to producing highly detailed or complex images based on textual descriptions. In order to explore these limitations, we propose a Structured Semantic Alignment (SSA) method for evaluating text-to-image generation models. SSA focuses on learning structured semantic embeddings across different modalities and aligning them in a joint space. The method employs the following steps to achieve its objective: (i) Generating mutated prompts by substituting words with semantically equivalent or nonequivalent alternatives while preserving the original syntax; (ii) Representing the sentence structure through parsing trees obtained via syntax parsing; (iii) Learning fine-grained structured embeddings that project semantic features from different modalities into a shared embedding space; (iv) Evaluating the semantic consistency between the structured text embeddings and the corresponding visual embeddings. Through experiments conducted on various benchmarks, we have demonstrated that SSA offers improved measurement of semantic consistency of text-to-image generation models. Additionally, it unveils a wide range of generation errors including under-generation, incorrect constituency, incorrect dependency, and semantic confusion. By uncovering these biases and limitations embedded within the models, our proposed method provides valuable insights into their shortcomings when applied to real-world scenarios.",
}
| Despite significant advancements in text-to-image generation models, they still face challenges when it comes to producing highly detailed or complex images based on textual descriptions. In order to explore these limitations, we propose a Structured Semantic Alignment (SSA) method for evaluating text-to-image generation models. SSA focuses on learning structured semantic embeddings across different modalities and aligning them in a joint space. The method employs the following steps to achieve its objective: (i) Generating mutated prompts by substituting words with semantically equivalent or nonequivalent alternatives while preserving the original syntax; (ii) Representing the sentence structure through parsing trees obtained via syntax parsing; (iii) Learning fine-grained structured embeddings that project semantic features from different modalities into a shared embedding space; (iv) Evaluating the semantic consistency between the structured text embeddings and the corresponding visual embeddings. Through experiments conducted on various benchmarks, we have demonstrated that SSA offers improved measurement of semantic consistency of text-to-image generation models. Additionally, it unveils a wide range of generation errors including under-generation, incorrect constituency, incorrect dependency, and semantic confusion. By uncovering these biases and limitations embedded within the models, our proposed method provides valuable insights into their shortcomings when applied to real-world scenarios. | [
"Feng, Qianyu",
"Sui, Yulei",
"Zhang, Hongyu"
] | Uncovering Limitations in Text-to-Image Generation: A Contrastive Approach with Structured Semantic Alignment | findings-emnlp.595 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.596.bib | https://aclanthology.org/2023.findings-emnlp.596/ | @inproceedings{shu-etal-2023-intent,
title = "An Intent-based and Annotation-free Method for Duplicate Question Detection in {CQA} Forums",
author = "Shu, Yubo and
Gu, Hansu and
Zhang, Peng and
Lu, Tun and
Gu, Ning",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.596",
doi = "10.18653/v1/2023.findings-emnlp.596",
pages = "8889--8899",
abstract = "With the advent of large language models (LLMs), Community Question Answering (CQA) forums offer well-curated questions and answers that can be utilized for instruction-tuning, effectively training LLMs to be aligned with human intents. However, the issue of duplicate questions arises as the volume of content within CQA continues to grow, posing a threat to content quality. Recent research highlights the benefits of detecting and eliminating duplicate content. It not only enhances the LLMs{'} ability to generalize across diverse intents but also improves the efficiency of training data utilization while addressing concerns related to information leakage. However, existing methods for detecting duplicate questions in CQA typically rely on generic text-pair matching models, overlooking the intent behind the questions. In this paper, we propose a novel intent-based duplication detector named Intent-DQD that comprehensively leverages intent information to address the problem of duplicate question detection in CQA. Intent-DQD first leverages the characteristics in CQA forums and extracts training labels to recognize and match intents without human annotation. Intent-DQD then effectively aggregates intent-level relations and establishes question-level relations to enable intent-aware duplication detection. Experimental results on fifteen distinct domains from both CQADupStack and Stack Overflow datasets demonstrate the effectiveness of Intent-DQD. Reproducible codes and datasets will be released upon publication of the paper.",
}
| With the advent of large language models (LLMs), Community Question Answering (CQA) forums offer well-curated questions and answers that can be utilized for instruction-tuning, effectively training LLMs to be aligned with human intents. However, the issue of duplicate questions arises as the volume of content within CQA continues to grow, posing a threat to content quality. Recent research highlights the benefits of detecting and eliminating duplicate content. It not only enhances the LLMs{'} ability to generalize across diverse intents but also improves the efficiency of training data utilization while addressing concerns related to information leakage. However, existing methods for detecting duplicate questions in CQA typically rely on generic text-pair matching models, overlooking the intent behind the questions. In this paper, we propose a novel intent-based duplication detector named Intent-DQD that comprehensively leverages intent information to address the problem of duplicate question detection in CQA. Intent-DQD first leverages the characteristics in CQA forums and extracts training labels to recognize and match intents without human annotation. Intent-DQD then effectively aggregates intent-level relations and establishes question-level relations to enable intent-aware duplication detection. Experimental results on fifteen distinct domains from both CQADupStack and Stack Overflow datasets demonstrate the effectiveness of Intent-DQD. Reproducible codes and datasets will be released upon publication of the paper. | [
"Shu, Yubo",
"Gu, Hansu",
"Zhang, Peng",
"Lu, Tun",
"Gu, Ning"
] | An Intent-based and Annotation-free Method for Duplicate Question Detection in CQA Forums | findings-emnlp.596 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.597.bib | https://aclanthology.org/2023.findings-emnlp.597/ | @inproceedings{cheng-etal-2023-accelerating,
title = "Accelerating Multiple Intent Detection and Slot Filling via Targeted Knowledge Distillation",
author = "Cheng, Xuxin and
Zhu, Zhihong and
Xu, Wanshi and
Li, Yaowei and
Li, Hongxiang and
Zou, Yuexian",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.597",
doi = "10.18653/v1/2023.findings-emnlp.597",
pages = "8900--8910",
abstract = "Recent non-autoregressive Spoken Language Understanding (SLU) models have attracted increasing attention because of their encouraging inference speed. However, most of existing methods (1) suffer from the multi-modality problem since they have little prior knowledge about the reference during inference; (2) fail to achieve a satisfactory inference speed limited by their complex frameworks. To tackle these issues, in this paper, we propose a $\textbf{T}$argeted $\textbf{K}$nowledge $\textbf{D}$istillation $\textbf{F}$ramework (TKDF) for multi-intent SLU, which utilizes the knowledge distillation method to improve the performance. Specifically, we first train an SLU model as the teacher model, which has higher accuracy while slower inference speed. Then we introduce an evaluator and apply a curriculum learning strategy to select proper targets for the student model. Experiment results on two public multi-intent datasets show that our approach can realize a flexible trade-off between inference speed and accuracy, achieving comparable performance to the state-of-the-art models while speeding up by over 4.5 times. More encouragingly, further analysis shows that distilling only 4{\%} of the original data can help the student model outperform its counterpart trained on the original data by about 14.6{\%} in terms of overall accuracy on MixATIS dataset.",
}
| Recent non-autoregressive Spoken Language Understanding (SLU) models have attracted increasing attention because of their encouraging inference speed. However, most of existing methods (1) suffer from the multi-modality problem since they have little prior knowledge about the reference during inference; (2) fail to achieve a satisfactory inference speed limited by their complex frameworks. To tackle these issues, in this paper, we propose a $\textbf{T}$argeted $\textbf{K}$nowledge $\textbf{D}$istillation $\textbf{F}$ramework (TKDF) for multi-intent SLU, which utilizes the knowledge distillation method to improve the performance. Specifically, we first train an SLU model as the teacher model, which has higher accuracy while slower inference speed. Then we introduce an evaluator and apply a curriculum learning strategy to select proper targets for the student model. Experiment results on two public multi-intent datasets show that our approach can realize a flexible trade-off between inference speed and accuracy, achieving comparable performance to the state-of-the-art models while speeding up by over 4.5 times. More encouragingly, further analysis shows that distilling only 4{\%} of the original data can help the student model outperform its counterpart trained on the original data by about 14.6{\%} in terms of overall accuracy on MixATIS dataset. | [
"Cheng, Xuxin",
"Zhu, Zhihong",
"Xu, Wanshi",
"Li, Yaowei",
"Li, Hongxiang",
"Zou, Yuexian"
] | Accelerating Multiple Intent Detection and Slot Filling via Targeted Knowledge Distillation | findings-emnlp.597 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.598.bib | https://aclanthology.org/2023.findings-emnlp.598/ | @inproceedings{li-etal-2023-type-aware,
title = "Type-Aware Decomposed Framework for Few-Shot Named Entity Recognition",
author = "Li, Yongqi and
Yu, Yu and
Qian, Tieyun",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.598",
doi = "10.18653/v1/2023.findings-emnlp.598",
pages = "8911--8927",
abstract = "Despite the recent success achieved by several two-stage prototypical networks in few-shot named entity recognition (NER) task, the over-detected false spans at span detection stage and the inaccurate and unstable prototypes at type classification stage remain to be challenging problems. In this paper, we propose a novel Type-Aware Decomposed framework, namely TadNER, to solve these problems. We first present a type-aware span filtering strategy to filter out false spans by removing those semantically far away from type names. We then present a type-aware contrastive learning strategy to construct more accurate and stable prototypes by jointly exploiting support samples and type names as references. Extensive experiments on various benchmarks prove that our proposed TadNER framework yields a new state-of-the-art performance.",
}
| Despite the recent success achieved by several two-stage prototypical networks in few-shot named entity recognition (NER) task, the over-detected false spans at span detection stage and the inaccurate and unstable prototypes at type classification stage remain to be challenging problems. In this paper, we propose a novel Type-Aware Decomposed framework, namely TadNER, to solve these problems. We first present a type-aware span filtering strategy to filter out false spans by removing those semantically far away from type names. We then present a type-aware contrastive learning strategy to construct more accurate and stable prototypes by jointly exploiting support samples and type names as references. Extensive experiments on various benchmarks prove that our proposed TadNER framework yields a new state-of-the-art performance. | [
"Li, Yongqi",
"Yu, Yu",
"Qian, Tieyun"
] | Type-Aware Decomposed Framework for Few-Shot Named Entity Recognition | findings-emnlp.598 | 2302.06397 | [
"https://github.com/liyongqi2002/TadNER"
] | https://huggingface.co/papers/2302.06397 | 0 | 0 | 0 | 2 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.599.bib | https://aclanthology.org/2023.findings-emnlp.599/ | @inproceedings{chiang-lee-2023-closer,
title = "A Closer Look into Using Large Language Models for Automatic Evaluation",
author = "Chiang, Cheng-Han and
Lee, Hung-yi",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.599",
doi = "10.18653/v1/2023.findings-emnlp.599",
pages = "8928--8942",
abstract = "Using large language models (LLMs) to evaluate text quality has recently gained popularity. Some existing prior works explore the idea of using LLMs for evaluation, while they differ in some details of the evaluation process. In this paper, we analyze *LLM evaluation* and *G-Eval*, and we discuss how those details in the evaluation process change how well the ratings given by LLMs correlate with human ratings. We find that the auto Chain-of-Thought (CoT) used in G-Eval does not always make G-Eval more aligned with human ratings. We also show that forcing the LLM to output only a numeric rating, as in G-Eval, is suboptimal. Last, we reveal that asking the LLM to explain its own ratings consistently improves the correlation between the ChatGPT and human ratings and pushes state-of-the-art (SoTA) correlations on two meta-evaluation datasets.",
}
| Using large language models (LLMs) to evaluate text quality has recently gained popularity. Some existing prior works explore the idea of using LLMs for evaluation, while they differ in some details of the evaluation process. In this paper, we analyze *LLM evaluation* and *G-Eval*, and we discuss how those details in the evaluation process change how well the ratings given by LLMs correlate with human ratings. We find that the auto Chain-of-Thought (CoT) used in G-Eval does not always make G-Eval more aligned with human ratings. We also show that forcing the LLM to output only a numeric rating, as in G-Eval, is suboptimal. Last, we reveal that asking the LLM to explain its own ratings consistently improves the correlation between the ChatGPT and human ratings and pushes state-of-the-art (SoTA) correlations on two meta-evaluation datasets. | [
"Chiang, Cheng-Han",
"Lee, Hung-yi"
] | A Closer Look into Using Large Language Models for Automatic Evaluation | findings-emnlp.599 | [
""
] | https://huggingface.co/papers/2310.05657 | 0 | 0 | 0 | 2 | [] | [] | [] | 1 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.600.bib | https://aclanthology.org/2023.findings-emnlp.600/ | @inproceedings{bugueno-de-melo-2023-connecting,
title = "Connecting the Dots: What Graph-Based Text Representations Work Best for Text Classification using Graph Neural Networks?",
author = "Bugue{\~n}o, Margarita and
de Melo, Gerard",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.600",
doi = "10.18653/v1/2023.findings-emnlp.600",
pages = "8943--8960",
abstract = "Given the success of Graph Neural Networks (GNNs) for structure-aware machine learning, many studies have explored their use for text classification, but mostly in specific domains with limited data characteristics. Moreover, some strategies prior to GNNs relied on graph mining and classical machine learning, making it difficult to assess their effectiveness in modern settings. This work extensively investigates graph representation methods for text classification, identifying practical implications and open challenges. We compare different graph construction schemes using a variety of GNN architectures and setups across five datasets, encompassing short and long documents as well as unbalanced scenarios in diverse domains. Two Transformer-based large language models are also included to complement the study. The results show that i) although the effectiveness of graphs depends on the textual input features and domain, simple graph constructions perform better the longer the documents are, ii) graph representations are especially beneficial for longer documents, outperforming Transformer-based models, iii) graph methods are particularly efficient for solving the task.",
}
| Given the success of Graph Neural Networks (GNNs) for structure-aware machine learning, many studies have explored their use for text classification, but mostly in specific domains with limited data characteristics. Moreover, some strategies prior to GNNs relied on graph mining and classical machine learning, making it difficult to assess their effectiveness in modern settings. This work extensively investigates graph representation methods for text classification, identifying practical implications and open challenges. We compare different graph construction schemes using a variety of GNN architectures and setups across five datasets, encompassing short and long documents as well as unbalanced scenarios in diverse domains. Two Transformer-based large language models are also included to complement the study. The results show that i) although the effectiveness of graphs depends on the textual input features and domain, simple graph constructions perform better the longer the documents are, ii) graph representations are especially beneficial for longer documents, outperforming Transformer-based models, iii) graph methods are particularly efficient for solving the task. | [
"Bugue{\\~n}o, Margarita",
"de Melo, Gerard"
] | Connecting the Dots: What Graph-Based Text Representations Work Best for Text Classification using Graph Neural Networks? | findings-emnlp.600 | 2305.14578 | [
"https://github.com/buguemar/grtc_gnns"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.601.bib | https://aclanthology.org/2023.findings-emnlp.601/ | @inproceedings{zocca-2023-natural,
title = "Natural Language Annotations for Reasoning about Program Semantics",
author = "Zocca, Marco",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.601",
doi = "10.18653/v1/2023.findings-emnlp.601",
pages = "8961--8966",
abstract = "By grounding natural language inference in code (and vice versa), researchers aim to create programming assistants that explain their work, are {``}coachable{''} and can surface any gaps in their reasoning. Can we deduce automatically interesting properties of programs from their syntax and common-sense annotations alone, without resorting to static analysis? How much of program logic and behaviour can be captured in natural language? To stimulate research in this direction and attempt to answer these questions we propose HTL, a dataset and protocol for annotating programs with natural language predicates at a finer granularity than code comments and without relying on internal compiler representations. The dataset is available at the following address: https://doi.org/10.5281/zenodo.7893113 .",
}
| By grounding natural language inference in code (and vice versa), researchers aim to create programming assistants that explain their work, are {``}coachable{''} and can surface any gaps in their reasoning. Can we deduce automatically interesting properties of programs from their syntax and common-sense annotations alone, without resorting to static analysis? How much of program logic and behaviour can be captured in natural language? To stimulate research in this direction and attempt to answer these questions we propose HTL, a dataset and protocol for annotating programs with natural language predicates at a finer granularity than code comments and without relying on internal compiler representations. The dataset is available at the following address: https://doi.org/10.5281/zenodo.7893113 . | [
"Zocca, Marco"
] | Natural Language Annotations for Reasoning about Program Semantics | findings-emnlp.601 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.602.bib | https://aclanthology.org/2023.findings-emnlp.602/ | @inproceedings{slaughter-etal-2023-pre,
title = "Pre-trained Speech Processing Models Contain Human-Like Biases that Propagate to Speech Emotion Recognition",
author = "Slaughter, Isaac and
Greenberg, Craig and
Schwartz, Reva and
Caliskan, Aylin",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.602",
doi = "10.18653/v1/2023.findings-emnlp.602",
pages = "8967--8989",
abstract = "Previous work has established that a person{'}s demographics and speech style affect how well speech processing models perform for them. But where does this bias come from? In this work, we present the Speech Embedding Association Test (SpEAT), a method for detecting bias in one type of model used for many speech tasks: pre-trained models. The SpEAT is inspired by word embedding association tests in natural language processing, which quantify intrinsic bias in a model{'}s representations of different concepts, such as race or valence{---}something{'}s pleasantness or unpleasantness{---}and capture the extent to which a model trained on large-scale socio-cultural data has learned human-like biases. Using the SpEAT, we test for six types of bias in 16 English speech models (including 4 models also trained on multilingual data), which come from the wav2vec 2.0, HuBERT, WavLM, and Whisper model families. We find that 14 or more models reveal positive valence (pleasantness) associations with abled people over disabled people, with European-Americans over African-Americans, with females over males, with U.S. accented speakers over non-U.S. accented speakers, and with younger people over older people. Beyond establishing that pre-trained speech models contain these biases, we also show that they can have real world effects. We compare biases found in pre-trained models to biases in downstream models adapted to the task of Speech Emotion Recognition (SER) and find that in 66 of the 96 tests performed (69{\%}), the group that is more associated with positive valence as indicated by the SpEAT also tends to be predicted as speaking with higher valence by the downstream model. Our work provides evidence that, like text and image-based models, pre-trained speech based-models frequently learn human-like biases when trained on large-scale socio-cultural datasets. Our work also shows that bias found in pre-trained models can propagate to the downstream task of SER.",
}
| Previous work has established that a person{'}s demographics and speech style affect how well speech processing models perform for them. But where does this bias come from? In this work, we present the Speech Embedding Association Test (SpEAT), a method for detecting bias in one type of model used for many speech tasks: pre-trained models. The SpEAT is inspired by word embedding association tests in natural language processing, which quantify intrinsic bias in a model{'}s representations of different concepts, such as race or valence{---}something{'}s pleasantness or unpleasantness{---}and capture the extent to which a model trained on large-scale socio-cultural data has learned human-like biases. Using the SpEAT, we test for six types of bias in 16 English speech models (including 4 models also trained on multilingual data), which come from the wav2vec 2.0, HuBERT, WavLM, and Whisper model families. We find that 14 or more models reveal positive valence (pleasantness) associations with abled people over disabled people, with European-Americans over African-Americans, with females over males, with U.S. accented speakers over non-U.S. accented speakers, and with younger people over older people. Beyond establishing that pre-trained speech models contain these biases, we also show that they can have real world effects. We compare biases found in pre-trained models to biases in downstream models adapted to the task of Speech Emotion Recognition (SER) and find that in 66 of the 96 tests performed (69{\%}), the group that is more associated with positive valence as indicated by the SpEAT also tends to be predicted as speaking with higher valence by the downstream model. Our work provides evidence that, like text and image-based models, pre-trained speech based-models frequently learn human-like biases when trained on large-scale socio-cultural datasets. Our work also shows that bias found in pre-trained models can propagate to the downstream task of SER. | [
"Slaughter, Isaac",
"Greenberg, Craig",
"Schwartz, Reva",
"Caliskan, Aylin"
] | Pre-trained Speech Processing Models Contain Human-Like Biases that Propagate to Speech Emotion Recognition | findings-emnlp.602 | 2310.18877 | [
"https://github.com/isaaconline/speat"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.603.bib | https://aclanthology.org/2023.findings-emnlp.603/ | @inproceedings{sun-etal-2023-text,
title = "Text Classification via Large Language Models",
author = "Sun, Xiaofei and
Li, Xiaoya and
Li, Jiwei and
Wu, Fei and
Guo, Shangwei and
Zhang, Tianwei and
Wang, Guoyin",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.603",
doi = "10.18653/v1/2023.findings-emnlp.603",
pages = "8990--9005",
abstract = "Despite the remarkable success of large-scale Language Models (LLMs) such as GPT-3, their performances still significantly underperform fine-tuned models in the task of text classification.This is due to (1) the lack of reasoning ability in addressing complex linguistic phenomena (e.g., intensification, contrast, irony etc); (2) limited number of tokens allowed in in-context learning. In this paper, we introduce \textbf{C}lue \textbf{A}nd \textbf{R}easoning \textbf{P}rompting (CARP). CARP adopts a progressive reasoning strategy tailored to addressing the complex linguistic phenomena involved in text classification: CARP first prompts LLMs to find superficial clues (e.g., keywords, tones, semantic relations, references, etc), based on which a diagnostic reasoning process is induced for final decisions. To further address the limited-token issue, CARP uses a fine-tuned model on the supervised dataset for $k$NN demonstration search in the in-context learning, allowing the model to take the advantage of both LLM{'}s generalization ability and the task-specific evidence provided by the full labeled dataset. Remarkably, CARP yields new SOTA performances on 4 out of 5 widely-used text-classification benchmarks, 97.39 (+1.24) on SST-2, 96.40 (+0.72) on AGNews, 98.78 (+0.25) on R8 and 96.95 (+0.6) on R52, and a performance comparable to SOTA on MR (92.39 v.s. 93.3). More importantly, we find that CARP delivers impressive abilities on low-resource and domain-adaptation setups. Specifically, using 16 examples per class, CARP achieves comparable performances to supervised models with 1,024 examples per class.",
}
| Despite the remarkable success of large-scale Language Models (LLMs) such as GPT-3, their performances still significantly underperform fine-tuned models in the task of text classification.This is due to (1) the lack of reasoning ability in addressing complex linguistic phenomena (e.g., intensification, contrast, irony etc); (2) limited number of tokens allowed in in-context learning. In this paper, we introduce \textbf{C}lue \textbf{A}nd \textbf{R}easoning \textbf{P}rompting (CARP). CARP adopts a progressive reasoning strategy tailored to addressing the complex linguistic phenomena involved in text classification: CARP first prompts LLMs to find superficial clues (e.g., keywords, tones, semantic relations, references, etc), based on which a diagnostic reasoning process is induced for final decisions. To further address the limited-token issue, CARP uses a fine-tuned model on the supervised dataset for $k$NN demonstration search in the in-context learning, allowing the model to take the advantage of both LLM{'}s generalization ability and the task-specific evidence provided by the full labeled dataset. Remarkably, CARP yields new SOTA performances on 4 out of 5 widely-used text-classification benchmarks, 97.39 (+1.24) on SST-2, 96.40 (+0.72) on AGNews, 98.78 (+0.25) on R8 and 96.95 (+0.6) on R52, and a performance comparable to SOTA on MR (92.39 v.s. 93.3). More importantly, we find that CARP delivers impressive abilities on low-resource and domain-adaptation setups. Specifically, using 16 examples per class, CARP achieves comparable performances to supervised models with 1,024 examples per class. | [
"Sun, Xiaofei",
"Li, Xiaoya",
"Li, Jiwei",
"Wu, Fei",
"Guo, Shangwei",
"Zhang, Tianwei",
"Wang, Guoyin"
] | Text Classification via Large Language Models | findings-emnlp.603 | 2305.08377 | [
"https://github.com/shannonai/gpt-cls-carp"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.604.bib | https://aclanthology.org/2023.findings-emnlp.604/ | @inproceedings{chen-etal-2023-task,
title = "On Task-personalized Multimodal Few-shot Learning for Visually-rich Document Entity Retrieval",
author = "Chen, Jiayi and
Dai, Hanjun and
Dai, Bo and
Zhang, Aidong and
Wei, Wei",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.604",
doi = "10.18653/v1/2023.findings-emnlp.604",
pages = "9006--9025",
abstract = "Visually-rich document entity retrieval (VDER), which extracts key information (e.g. date, address) from document images like invoices and receipts, has become an important topic in industrial NLP applications. The emergence of new document types at a constant pace, each with its unique entity types, presents a unique challenge: many documents contain unseen entity types that occur only a couple of times. Addressing this challenge requires models to have the ability of learning entities in a few-shot manner. However, prior works for Few-shot VDER mainly address the problem at the document level with a predefined global entity space, which doesn{'}t account for the entity-level few-shot scenario: target entity types are locally personalized by each task and entity occurrences vary significantly among documents. To address this unexplored scenario, this paper studies a novel entity-level few-shot VDER task. The challenges lie in the uniqueness of the label space for each task and the increased complexity of out-of-distribution (OOD) contents. To tackle this novel task, we present a task-aware meta-learning based framework, with a central focus on achieving effective task personalization that distinguishes between in-task and out-of-task distribution. Specifically, we adopt a hierarchical decoder (HC) and employ contrastive learning (ContrastProtoNet) to achieve this goal. Furthermore, we introduce a new dataset, FewVEX, to boost future research in the field of entity-level few-shot VDER. Experimental results demonstrate our approaches significantly improve the robustness of popular meta-learning baselines.",
}
| Visually-rich document entity retrieval (VDER), which extracts key information (e.g. date, address) from document images like invoices and receipts, has become an important topic in industrial NLP applications. The emergence of new document types at a constant pace, each with its unique entity types, presents a unique challenge: many documents contain unseen entity types that occur only a couple of times. Addressing this challenge requires models to have the ability of learning entities in a few-shot manner. However, prior works for Few-shot VDER mainly address the problem at the document level with a predefined global entity space, which doesn{'}t account for the entity-level few-shot scenario: target entity types are locally personalized by each task and entity occurrences vary significantly among documents. To address this unexplored scenario, this paper studies a novel entity-level few-shot VDER task. The challenges lie in the uniqueness of the label space for each task and the increased complexity of out-of-distribution (OOD) contents. To tackle this novel task, we present a task-aware meta-learning based framework, with a central focus on achieving effective task personalization that distinguishes between in-task and out-of-task distribution. Specifically, we adopt a hierarchical decoder (HC) and employ contrastive learning (ContrastProtoNet) to achieve this goal. Furthermore, we introduce a new dataset, FewVEX, to boost future research in the field of entity-level few-shot VDER. Experimental results demonstrate our approaches significantly improve the robustness of popular meta-learning baselines. | [
"Chen, Jiayi",
"Dai, Hanjun",
"Dai, Bo",
"Zhang, Aidong",
"Wei, Wei"
] | On Task-personalized Multimodal Few-shot Learning for Visually-rich Document Entity Retrieval | findings-emnlp.604 | 2311.00693 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.605.bib | https://aclanthology.org/2023.findings-emnlp.605/ | @inproceedings{murthy-etal-2023-semi,
title = "Semi-Structured Object Sequence Encoders",
author = "Murthy, Rudra and
Bhat, Riyaz and
Gunasekara, Chulaka and
Patel, Siva and
Wan, Hui and
Dhamecha, Tejas and
Contractor, Danish and
Danilevsky, Marina",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.605",
doi = "10.18653/v1/2023.findings-emnlp.605",
pages = "9026--9039",
abstract = "In this paper we explore the task of modeling semi-structured object sequences; in particular, we focus our attention on the problem of developing a structure-aware input representation for such sequences. Examples of such data include user activity on websites, machine logs, and many others. This type of data is often represented as a sequence of sets of key-value pairs over time and can present modeling challenges due to an ever-increasing sequence length. We propose a two-part approach, which first considers each key independently and encodes a representation of its values over time; we then self-attend over these value-aware key representations to accomplish a downstream task. This allows us to operate on longer object sequences than existing methods. We introduce a novel shared-attention-head architecture between the two modules and present an innovative training schedule that interleaves the training of both modules with shared weights for some attention heads. Our experiments on multiple prediction tasks using real-world data demonstrate that our approach outperforms a unified network with hierarchical encoding, as well as other methods including a \textit{record-centric} representation and a \textit{flattened} representation of the sequence.",
}
| In this paper we explore the task of modeling semi-structured object sequences; in particular, we focus our attention on the problem of developing a structure-aware input representation for such sequences. Examples of such data include user activity on websites, machine logs, and many others. This type of data is often represented as a sequence of sets of key-value pairs over time and can present modeling challenges due to an ever-increasing sequence length. We propose a two-part approach, which first considers each key independently and encodes a representation of its values over time; we then self-attend over these value-aware key representations to accomplish a downstream task. This allows us to operate on longer object sequences than existing methods. We introduce a novel shared-attention-head architecture between the two modules and present an innovative training schedule that interleaves the training of both modules with shared weights for some attention heads. Our experiments on multiple prediction tasks using real-world data demonstrate that our approach outperforms a unified network with hierarchical encoding, as well as other methods including a \textit{record-centric} representation and a \textit{flattened} representation of the sequence. | [
"Murthy, Rudra",
"Bhat, Riyaz",
"Gunasekara, Chulaka",
"Patel, Siva",
"Wan, Hui",
"Dhamecha, Tejas",
"Contractor, Danish",
"Danilevsky, Marina"
] | Semi-Structured Object Sequence Encoders | findings-emnlp.605 | 2301.01015 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.606.bib | https://aclanthology.org/2023.findings-emnlp.606/ | @inproceedings{xu-etal-2023-detime,
title = "{D}e{T}i{ME}: Diffusion-Enhanced Topic Modeling using Encoder-decoder based {LLM}",
author = "Xu, Weijie and
Hu, Wenxiang and
Wu, Fanyou and
Sengamedu, Srinivasan",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.606",
doi = "10.18653/v1/2023.findings-emnlp.606",
pages = "9040--9057",
abstract = "In the burgeoning field of natural language processing, Neural Topic Models (NTMs) and Large Language Models (LLMs) have emerged as areas of significant research interest. Despite this, NTMs primarily utilize contextual embeddings from LLMs, which are not optimal for clustering or capable for topic generation. Our study addresses this gap by introducing a novel framework named Diffusion-Enhanced Topic Modeling using Encoder-Decoder-based LLMs (DeTiME). DeTiME leverages Encoder-Decoder-based LLMs to produce highly clusterable embeddings that could generate topics that exhibit both superior clusterability and enhanced semantic coherence compared to existing methods. Additionally, by exploiting the power of diffusion, our framework also provides the capability to generate content relevant to the identified topics. This dual functionality allows users to efficiently produce highly clustered topics and related content simultaneously. DeTiME{'}s potential extends to generating clustered embeddings as well. Notably, our proposed framework proves to be efficient to train and exhibits high adaptability, demonstrating its potential for a wide array of applications.",
}
| In the burgeoning field of natural language processing, Neural Topic Models (NTMs) and Large Language Models (LLMs) have emerged as areas of significant research interest. Despite this, NTMs primarily utilize contextual embeddings from LLMs, which are not optimal for clustering or capable for topic generation. Our study addresses this gap by introducing a novel framework named Diffusion-Enhanced Topic Modeling using Encoder-Decoder-based LLMs (DeTiME). DeTiME leverages Encoder-Decoder-based LLMs to produce highly clusterable embeddings that could generate topics that exhibit both superior clusterability and enhanced semantic coherence compared to existing methods. Additionally, by exploiting the power of diffusion, our framework also provides the capability to generate content relevant to the identified topics. This dual functionality allows users to efficiently produce highly clustered topics and related content simultaneously. DeTiME{'}s potential extends to generating clustered embeddings as well. Notably, our proposed framework proves to be efficient to train and exhibits high adaptability, demonstrating its potential for a wide array of applications. | [
"Xu, Weijie",
"Hu, Wenxiang",
"Wu, Fanyou",
"Sengamedu, Srinivasan"
] | DeTiME: Diffusion-Enhanced Topic Modeling using Encoder-decoder based LLM | findings-emnlp.606 | 2310.15296 | [
"https://github.com/amazon-science/text_generation_diffusion_llm_topic"
] | https://huggingface.co/papers/2310.15296 | 1 | 3 | 0 | 4 | [] | [
"xwjzds/paraphrase_collections",
"xwjzds/pretrain_sts_long",
"xwjzds/pretrain_sts_similarity"
] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.607.bib | https://aclanthology.org/2023.findings-emnlp.607/ | @inproceedings{wang-etal-2023-energy,
title = "Energy and Carbon Considerations of Fine-Tuning {BERT}",
author = "Wang, Xiaorong and
Na, Clara and
Strubell, Emma and
Friedler, Sorelle and
Luccioni, Sasha",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.607",
doi = "10.18653/v1/2023.findings-emnlp.607",
pages = "9058--9069",
abstract = "Despite the popularity of the pre-train then fine-tune paradigm in the NLP community, existing work quantifying energy costs and associated carbon emissions has largely focused on language model pre-training. Although a single pre-training run draws substantially more energy than fine-tuning, fine-tuning is performed more frequently by many more individual actors, and thus must be accounted for when considering the energy and carbon footprint of NLP. In order to better characterize the role of fine-tuning in the landscape of energy and carbon emissions in NLP, we perform a careful empirical study of the computational costs of fine-tuning across tasks, datasets, hardware infrastructure and measurement modalities. Our experimental results allow us to place fine-tuning energy and carbon costs into perspective with respect to pre-training and inference, and outline recommendations to NLP researchers and practitioners who wish to improve their fine-tuning energy efficiency.",
}
| Despite the popularity of the pre-train then fine-tune paradigm in the NLP community, existing work quantifying energy costs and associated carbon emissions has largely focused on language model pre-training. Although a single pre-training run draws substantially more energy than fine-tuning, fine-tuning is performed more frequently by many more individual actors, and thus must be accounted for when considering the energy and carbon footprint of NLP. In order to better characterize the role of fine-tuning in the landscape of energy and carbon emissions in NLP, we perform a careful empirical study of the computational costs of fine-tuning across tasks, datasets, hardware infrastructure and measurement modalities. Our experimental results allow us to place fine-tuning energy and carbon costs into perspective with respect to pre-training and inference, and outline recommendations to NLP researchers and practitioners who wish to improve their fine-tuning energy efficiency. | [
"Wang, Xiaorong",
"Na, Clara",
"Strubell, Emma",
"Friedler, Sorelle",
"Luccioni, Sasha"
] | Energy and Carbon Considerations of Fine-Tuning BERT | findings-emnlp.607 | 2311.10267 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.608.bib | https://aclanthology.org/2023.findings-emnlp.608/ | @inproceedings{shashidhar-etal-2023-democratizing,
title = "Democratizing {LLM}s: An Exploration of Cost-Performance Trade-offs in Self-Refined Open-Source Models",
author = "Shashidhar, Sumuk and
Chinta, Abhinav and
Sahai, Vaibhav and
Wang, Zhenhailong and
Ji, Heng",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.608",
doi = "10.18653/v1/2023.findings-emnlp.608",
pages = "9070--9084",
abstract = "The dominance of proprietary LLMs has led to restricted access and raised information privacy concerns. The SoTA open-source alternatives are crucial for information-sensitive and high-volume applications but often lag behind in performance. To address this gap, we propose (1) A generalized variant of iterative self-critique and self-refinement devoid of external influence. (2) A novel ranking metric - Performance, Refinement, and Inference Cost Score (PeRFICS) - to find the optimal model for a given task considering refined performance and cost. Our experiments show that SoTA open source models of varying sizes from 7B - 65B, on average, improve 8.2{\%} from their baseline performance. Strikingly, even models with extremely small memory footprints, such as Vicuna-7B, show a 11.74{\%} improvement overall and up to a 25.39{\%} improvement in high-creativity, open ended tasks on the Vicuna benchmark. Vicuna-13B takes it a step further and outperforms ChatGPT post-refinement. This work has profound implications for resource-constrained and information-sensitive environments seeking to leverage LLMs without incurring prohibitive costs, compromising on performance and privacy. The domain-agnostic self-refinement process coupled with our novel ranking metric facilitates informed decision-making in model selection, thereby reducing costs and democratizing access to high-performing language models, as evidenced by three case studies on personal computing, gaming and enterprise solutions.",
}
| The dominance of proprietary LLMs has led to restricted access and raised information privacy concerns. The SoTA open-source alternatives are crucial for information-sensitive and high-volume applications but often lag behind in performance. To address this gap, we propose (1) A generalized variant of iterative self-critique and self-refinement devoid of external influence. (2) A novel ranking metric - Performance, Refinement, and Inference Cost Score (PeRFICS) - to find the optimal model for a given task considering refined performance and cost. Our experiments show that SoTA open source models of varying sizes from 7B - 65B, on average, improve 8.2{\%} from their baseline performance. Strikingly, even models with extremely small memory footprints, such as Vicuna-7B, show a 11.74{\%} improvement overall and up to a 25.39{\%} improvement in high-creativity, open ended tasks on the Vicuna benchmark. Vicuna-13B takes it a step further and outperforms ChatGPT post-refinement. This work has profound implications for resource-constrained and information-sensitive environments seeking to leverage LLMs without incurring prohibitive costs, compromising on performance and privacy. The domain-agnostic self-refinement process coupled with our novel ranking metric facilitates informed decision-making in model selection, thereby reducing costs and democratizing access to high-performing language models, as evidenced by three case studies on personal computing, gaming and enterprise solutions. | [
"Shashidhar, Sumuk",
"Chinta, Abhinav",
"Sahai, Vaibhav",
"Wang, Zhenhailong",
"Ji, Heng"
] | Democratizing LLMs: An Exploration of Cost-Performance Trade-offs in Self-Refined Open-Source Models | findings-emnlp.608 | 2310.07611 | [
""
] | https://huggingface.co/papers/2310.07611 | 2 | 2 | 0 | 5 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.609.bib | https://aclanthology.org/2023.findings-emnlp.609/ | @inproceedings{chen-etal-2023-chinese,
title = "{C}hinese Metaphorical Relation Extraction: Dataset and Models",
author = "Chen, Guihua and
Wu, Tiantian and
Cheng, MiaoMiao and
Han, Xu and
Gong, Jiefu and
Wang, Shijin and
Song, Wei",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.609",
doi = "10.18653/v1/2023.findings-emnlp.609",
pages = "9085--9095",
abstract = "Metaphor identification is usually formulated as a sequence labeling or a syntactically related word-pair classification problem. In this paper, we propose a novel formulation of metaphor identification as a relation extraction problem. We introduce metaphorical relations, which are links between two spans, a target span and a source-related span, which are realized in sentences. Based on spans, we can use more flexible and precise text units beyond single words for capturing the properties of the target and the source. We create a dataset for Chinese metaphorical relation extraction, with more than 4,200 sentences annotated with metaphorical relations, corresponding target/source-related spans, and fine-grained span types. We develop a span-based end-to-end model for metaphorical relation extraction and demonstrate its effectiveness. We expect that metaphorical relation extraction can serve as a bridge for connecting linguistic and conceptual metaphor processing. The dataset is at https://github.com/cnunlp/CMRE.",
}
| Metaphor identification is usually formulated as a sequence labeling or a syntactically related word-pair classification problem. In this paper, we propose a novel formulation of metaphor identification as a relation extraction problem. We introduce metaphorical relations, which are links between two spans, a target span and a source-related span, which are realized in sentences. Based on spans, we can use more flexible and precise text units beyond single words for capturing the properties of the target and the source. We create a dataset for Chinese metaphorical relation extraction, with more than 4,200 sentences annotated with metaphorical relations, corresponding target/source-related spans, and fine-grained span types. We develop a span-based end-to-end model for metaphorical relation extraction and demonstrate its effectiveness. We expect that metaphorical relation extraction can serve as a bridge for connecting linguistic and conceptual metaphor processing. The dataset is at https://github.com/cnunlp/CMRE. | [
"Chen, Guihua",
"Wu, Tiantian",
"Cheng, MiaoMiao",
"Han, Xu",
"Gong, Jiefu",
"Wang, Shijin",
"Song, Wei"
] | Chinese Metaphorical Relation Extraction: Dataset and Models | findings-emnlp.609 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.610.bib | https://aclanthology.org/2023.findings-emnlp.610/ | @inproceedings{volk-etal-2023-example,
title = "Example-based Hypernetworks for Multi-source Adaptation to Unseen Domains",
author = "Volk, Tomer and
Ben-David, Eyal and
Amosy, Ohad and
Chechik, Gal and
Reichart, Roi",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.610",
doi = "10.18653/v1/2023.findings-emnlp.610",
pages = "9096--9113",
abstract = "As Natural Language Processing (NLP) algorithms continually achieve new milestones, out-of-distribution generalization remains a significant challenge. This paper addresses the issue of multi-source adaptation for unfamiliar domains: We leverage labeled data from multiple source domains to generalize to unknown target domains at training. Our innovative framework employs example-based Hypernetwork adaptation: a T5 encoder-decoder initially generates a unique signature from an input example, embedding it within the source domains{'} semantic space. This signature is subsequently utilized by a Hypernetwork to generate the task classifier{'}s weights. In an advanced version, the signature also enriches the input example{'}s representation. We evaluated our method across two tasks{---}sentiment classification and natural language inference{---}in 29 adaptation scenarios, where it outpaced established algorithms. We also compare our finetuned architecture to few-shot GPT-3, demonstrating its effectiveness in essential use cases. To the best of our knowledge, this marks the first application of Hypernetworks to the adaptation for unknown domains.",
}
| As Natural Language Processing (NLP) algorithms continually achieve new milestones, out-of-distribution generalization remains a significant challenge. This paper addresses the issue of multi-source adaptation for unfamiliar domains: We leverage labeled data from multiple source domains to generalize to unknown target domains at training. Our innovative framework employs example-based Hypernetwork adaptation: a T5 encoder-decoder initially generates a unique signature from an input example, embedding it within the source domains{'} semantic space. This signature is subsequently utilized by a Hypernetwork to generate the task classifier{'}s weights. In an advanced version, the signature also enriches the input example{'}s representation. We evaluated our method across two tasks{---}sentiment classification and natural language inference{---}in 29 adaptation scenarios, where it outpaced established algorithms. We also compare our finetuned architecture to few-shot GPT-3, demonstrating its effectiveness in essential use cases. To the best of our knowledge, this marks the first application of Hypernetworks to the adaptation for unknown domains. | [
"Volk, Tomer",
"Ben-David, Eyal",
"Amosy, Ohad",
"Chechik, Gal",
"Reichart, Roi"
] | Example-based Hypernetworks for Multi-source Adaptation to Unseen Domains | findings-emnlp.610 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.611.bib | https://aclanthology.org/2023.findings-emnlp.611/ | @inproceedings{lin-etal-2023-beneath,
title = "Beneath the Surface: Unveiling Harmful Memes with Multimodal Reasoning Distilled from Large Language Models",
author = "Lin, Hongzhan and
Luo, Ziyang and
Ma, Jing and
Chen, Long",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.611",
doi = "10.18653/v1/2023.findings-emnlp.611",
pages = "9114--9128",
abstract = "The age of social media is rife with memes. Understanding and detecting harmful memes pose a significant challenge due to their implicit meaning that is not explicitly conveyed through the surface text and image. However, existing harmful meme detection approaches only recognize superficial harm-indicative signals in an end-to-end classification manner but ignore in-depth cognition of the meme text and image. In this paper, we attempt to detect harmful memes based on advanced reasoning over the interplay of multimodal information in memes. Inspired by the success of Large Language Models (LLMs) on complex reasoning, we first conduct abductive reasoning with LLMs. Then we propose a novel generative framework to learn reasonable thoughts from LLMs for better multimodal fusion and lightweight fine-tuning, which consists of two training stages: 1) Distill multimodal reasoning knowledge from LLMs; and 2) Fine-tune the generative framework to infer harmfulness. Extensive experiments conducted on three meme datasets demonstrate that our proposed approach achieves superior performance than state-of-the-art methods on the harmful meme detection task.",
}
| The age of social media is rife with memes. Understanding and detecting harmful memes pose a significant challenge due to their implicit meaning that is not explicitly conveyed through the surface text and image. However, existing harmful meme detection approaches only recognize superficial harm-indicative signals in an end-to-end classification manner but ignore in-depth cognition of the meme text and image. In this paper, we attempt to detect harmful memes based on advanced reasoning over the interplay of multimodal information in memes. Inspired by the success of Large Language Models (LLMs) on complex reasoning, we first conduct abductive reasoning with LLMs. Then we propose a novel generative framework to learn reasonable thoughts from LLMs for better multimodal fusion and lightweight fine-tuning, which consists of two training stages: 1) Distill multimodal reasoning knowledge from LLMs; and 2) Fine-tune the generative framework to infer harmfulness. Extensive experiments conducted on three meme datasets demonstrate that our proposed approach achieves superior performance than state-of-the-art methods on the harmful meme detection task. | [
"Lin, Hongzhan",
"Luo, Ziyang",
"Ma, Jing",
"Chen, Long"
] | Beneath the Surface: Unveiling Harmful Memes with Multimodal Reasoning Distilled from Large Language Models | findings-emnlp.611 | 2312.05434 | [
"https://github.com/hkbunlp/mr.harm-emnlp2023"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.612.bib | https://aclanthology.org/2023.findings-emnlp.612/ | @inproceedings{wang-etal-2023-domain,
title = "Domain Adaptation for Conversational Query Production with the {RAG} Model Feedback",
author = "Wang, Ante and
Song, Linfeng and
Xu, Ge and
Su, Jinsong",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.612",
doi = "10.18653/v1/2023.findings-emnlp.612",
pages = "9129--9141",
abstract = "Conversational query production is an emerging fundamental task for the dialogue system, where search queries are generated to explore the vast and continually updating knowledge from a search engine. To accelerate this line of research, previous studies have released several datasets with human-annotated search queries. However, the limited annotations still can not cover conversations of various domains. To solve this challenge, we propose a novel domain adaptation framework. It is inspired by a weakly supervised learning algorithm from previous work that guides a model using reinforcement learning with BM25 scores as feedback. Though effective, it is fragile facing noisy content on webpages from a commercial search engine and variance in conversations because of ignoring deep semantic information of dialogue contexts. Thus, we improve the algorithm by taking the advance of retrieval-augmented generation (RAG) and exploring several practical techniques such as knowledge distillation for stable training. We conduct experiments in multiple settings across different languages. Guided by the RAG model feedback, our model is more robust and performs significantly better especially in a more challenging setting over strong baselines.",
}
| Conversational query production is an emerging fundamental task for the dialogue system, where search queries are generated to explore the vast and continually updating knowledge from a search engine. To accelerate this line of research, previous studies have released several datasets with human-annotated search queries. However, the limited annotations still can not cover conversations of various domains. To solve this challenge, we propose a novel domain adaptation framework. It is inspired by a weakly supervised learning algorithm from previous work that guides a model using reinforcement learning with BM25 scores as feedback. Though effective, it is fragile facing noisy content on webpages from a commercial search engine and variance in conversations because of ignoring deep semantic information of dialogue contexts. Thus, we improve the algorithm by taking the advance of retrieval-augmented generation (RAG) and exploring several practical techniques such as knowledge distillation for stable training. We conduct experiments in multiple settings across different languages. Guided by the RAG model feedback, our model is more robust and performs significantly better especially in a more challenging setting over strong baselines. | [
"Wang, Ante",
"Song, Linfeng",
"Xu, Ge",
"Su, Jinsong"
] | Domain Adaptation for Conversational Query Production with the RAG Model Feedback | findings-emnlp.612 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.613.bib | https://aclanthology.org/2023.findings-emnlp.613/ | @inproceedings{he-etal-2023-lego,
title = "{LEGO}: A Multi-agent Collaborative Framework with Role-playing and Iterative Feedback for Causality Explanation Generation",
author = "He, Zhitao and
Cao, Pengfei and
Chen, Yubo and
Liu, Kang and
Li, Ruopeng and
Sun, Mengshu and
Zhao, Jun",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.613",
doi = "10.18653/v1/2023.findings-emnlp.613",
pages = "9142--9163",
abstract = "Causality Explanation Generation refers to generate an explanation in natural language given an initial cause-effect pair. It demands rigorous explicit rationales to demonstrate the acquisition of implicit commonsense knowledge, which is unlikely to be easily memorized, making it challenging for large language models since they are often suffering from spurious causal associations when they encounter the content that does not exist in their memory. In this work, we introduce LEGO, a Multi-agent Collaborative Framework with Role-playing and Iterative Feedback for causality explanation generation. Specifically, we treat LLM as character malleable LEGO block and utilize role-playing to assign specific roles to five LLMs. We firstly devise a Fine-grained World Knowledge Integration Module to augment information about tasks for alleviating the phenomenon of spurious causal associations. Then, we leverage an Iterative Feedback and Refinement Module to improve the generated explanation by multi-aspect feedback. Extensive experiments on widely used WIKIWHY and e-CARE datasets show the superiority of our multi-agent framework in terms of reasoning about the causality among cause and effect.",
}
| Causality Explanation Generation refers to generate an explanation in natural language given an initial cause-effect pair. It demands rigorous explicit rationales to demonstrate the acquisition of implicit commonsense knowledge, which is unlikely to be easily memorized, making it challenging for large language models since they are often suffering from spurious causal associations when they encounter the content that does not exist in their memory. In this work, we introduce LEGO, a Multi-agent Collaborative Framework with Role-playing and Iterative Feedback for causality explanation generation. Specifically, we treat LLM as character malleable LEGO block and utilize role-playing to assign specific roles to five LLMs. We firstly devise a Fine-grained World Knowledge Integration Module to augment information about tasks for alleviating the phenomenon of spurious causal associations. Then, we leverage an Iterative Feedback and Refinement Module to improve the generated explanation by multi-aspect feedback. Extensive experiments on widely used WIKIWHY and e-CARE datasets show the superiority of our multi-agent framework in terms of reasoning about the causality among cause and effect. | [
"He, Zhitao",
"Cao, Pengfei",
"Chen, Yubo",
"Liu, Kang",
"Li, Ruopeng",
"Sun, Mengshu",
"Zhao, Jun"
] | LEGO: A Multi-agent Collaborative Framework with Role-playing and Iterative Feedback for Causality Explanation Generation | findings-emnlp.613 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.614.bib | https://aclanthology.org/2023.findings-emnlp.614/ | @inproceedings{chakraborty-etal-2023-ranking,
title = "Ranking {LLM}-Generated Loop Invariants for Program Verification",
author = "Chakraborty, Saikat and
Lahiri, Shuvendu and
Fakhoury, Sarah and
Lal, Akash and
Musuvathi, Madanlal and
Rastogi, Aseem and
Senthilnathan, Aditya and
Sharma, Rahul and
Swamy, Nikhil",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.614",
doi = "10.18653/v1/2023.findings-emnlp.614",
pages = "9164--9175",
abstract = "Synthesizing inductive loop invariants is fundamental to automating program verification. In this work we observe that Large Language Models (such as gpt-3.5 or gpt-4) are capable of synthesizing loop invariants for a class of programs in a 0-shot setting, yet require several samples to generate the correct invariants. This can lead to a large number a calls to a program verifier to establish an invariant. To address this issue, we propose a re-ranking approach for the generated results of LLMs. We have designed a ranker that can distinguish between correct inductive invariants and incorrect attempts based on the problem definition. The ranker is optimized as a contrastive ranker. Experimental results demonstrate that this re-ranking mechanism significantly improves the ranking of correct invariants among the generated candidates, leading to a notable reduction in the number of calls to a verifier.",
}
| Synthesizing inductive loop invariants is fundamental to automating program verification. In this work we observe that Large Language Models (such as gpt-3.5 or gpt-4) are capable of synthesizing loop invariants for a class of programs in a 0-shot setting, yet require several samples to generate the correct invariants. This can lead to a large number a calls to a program verifier to establish an invariant. To address this issue, we propose a re-ranking approach for the generated results of LLMs. We have designed a ranker that can distinguish between correct inductive invariants and incorrect attempts based on the problem definition. The ranker is optimized as a contrastive ranker. Experimental results demonstrate that this re-ranking mechanism significantly improves the ranking of correct invariants among the generated candidates, leading to a notable reduction in the number of calls to a verifier. | [
"Chakraborty, Saikat",
"Lahiri, Shuvendu",
"Fakhoury, Sarah",
"Lal, Akash",
"Musuvathi, Madanlal",
"Rastogi, Aseem",
"Senthilnathan, Aditya",
"Sharma, Rahul",
"Swamy, Nikhil"
] | Ranking LLM-Generated Loop Invariants for Program Verification | findings-emnlp.614 | 2310.09342 | [
"https://github.com/microsoft/NeuralInvariantRanker"
] | https://huggingface.co/papers/2310.09342 | 3 | 2 | 1 | 9 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.findings-emnlp.615.bib | https://aclanthology.org/2023.findings-emnlp.615/ | @inproceedings{renner-etal-2023-wordnet,
title = "{W}ord{N}et Is All You Need: A Surprisingly Effective Unsupervised Method for Graded Lexical Entailment",
author = "Renner, Joseph and
Denis, Pascal and
Gilleron, R{\'e}mi",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.615",
doi = "10.18653/v1/2023.findings-emnlp.615",
pages = "9176--9182",
abstract = "We propose a simple unsupervised approach which exclusively relies on WordNet (Miller,1995) for predicting graded lexical entailment (GLE) in English. Inspired by the seminal work of Resnik (1995), our method models GLE as the sum of two information-theoretic scores: a symmetric semantic similarity score and an asymmetric specificity loss score, both exploiting the hierarchical synset structure of WordNet. Our approach also includes a simple disambiguation mechanism to handle polysemy in a given word pair. Despite its simplicity, our method achieves performance above the state of the art (Spearman $\rho$ = 0.75) on HyperLex (Vulic et al., 2017), the largest GLE dataset, outperforming all previous methods, including specialized word embeddings approaches that use WordNet as weak supervision.",
}
| We propose a simple unsupervised approach which exclusively relies on WordNet (Miller,1995) for predicting graded lexical entailment (GLE) in English. Inspired by the seminal work of Resnik (1995), our method models GLE as the sum of two information-theoretic scores: a symmetric semantic similarity score and an asymmetric specificity loss score, both exploiting the hierarchical synset structure of WordNet. Our approach also includes a simple disambiguation mechanism to handle polysemy in a given word pair. Despite its simplicity, our method achieves performance above the state of the art (Spearman $\rho$ = 0.75) on HyperLex (Vulic et al., 2017), the largest GLE dataset, outperforming all previous methods, including specialized word embeddings approaches that use WordNet as weak supervision. | [
"Renner, Joseph",
"Denis, Pascal",
"Gilleron, R{\\'e}mi"
] | WordNet Is All You Need: A Surprisingly Effective Unsupervised Method for Graded Lexical Entailment | findings-emnlp.615 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.findings-emnlp.616.bib | https://aclanthology.org/2023.findings-emnlp.616/ | @inproceedings{lee-etal-2023-knowledge,
title = "Knowledge Corpus Error in Question Answering",
author = "Lee, Yejoon and
Oh, Philhoon and
Thorne, James",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.616",
doi = "10.18653/v1/2023.findings-emnlp.616",
pages = "9183--9197",
abstract = "Recent works in open-domain question answering (QA) have explored generating context passages from large language models (LLMs), replacing the traditional retrieval step in the QA pipeline. However, it is not well understood why generated passages can be more effective than retrieved ones. This study revisits the conventional formulation of QA and introduces the concept of $\textit{knowledge corpus error}$. This error arises when the knowledge corpus used for retrieval is only a subset of the entire string space, potentially excluding more helpful passages that exist outside the corpus. LLMs may mitigate this shortcoming by generating passages in a larger space. We come up with an experiment of paraphrasing human-annotated gold context using LLMs to observe knowledge corpus error empirically. Our results across three QA benchmarks reveal an increased performance (10{\%} - 13{\%}) when using paraphrased passage, indicating a signal for the existence of knowledge corpus error.",
}
| Recent works in open-domain question answering (QA) have explored generating context passages from large language models (LLMs), replacing the traditional retrieval step in the QA pipeline. However, it is not well understood why generated passages can be more effective than retrieved ones. This study revisits the conventional formulation of QA and introduces the concept of $\textit{knowledge corpus error}$. This error arises when the knowledge corpus used for retrieval is only a subset of the entire string space, potentially excluding more helpful passages that exist outside the corpus. LLMs may mitigate this shortcoming by generating passages in a larger space. We come up with an experiment of paraphrasing human-annotated gold context using LLMs to observe knowledge corpus error empirically. Our results across three QA benchmarks reveal an increased performance (10{\%} - 13{\%}) when using paraphrased passage, indicating a signal for the existence of knowledge corpus error. | [
"Lee, Yejoon",
"Oh, Philhoon",
"Thorne, James"
] | Knowledge Corpus Error in Question Answering | findings-emnlp.616 | 2310.18076 | [
"https://github.com/xfactlab/emnlp2023-knowledge-corpus-error"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.617.bib | https://aclanthology.org/2023.findings-emnlp.617/ | @inproceedings{freitag-etal-2023-epsilon,
title = "Epsilon Sampling Rocks: Investigating Sampling Strategies for Minimum {B}ayes Risk Decoding for Machine Translation",
author = "Freitag, Markus and
Ghorbani, Behrooz and
Fernandes, Patrick",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.617",
doi = "10.18653/v1/2023.findings-emnlp.617",
pages = "9198--9209",
abstract = "Recent advances in machine translation (MT) have shown that Minimum Bayes Risk (MBR) decoding can be a powerful alternative to beam search decoding, especially when combined with neural-based utility functions. However, the performance of MBR decoding depends heavily on how and how many candidates are sampled from the model. In this paper, we explore how different sampling approaches for generating candidate lists for MBR decoding affect performance. We evaluate popular sampling approaches, such as ancestral, nucleus, and top-k sampling. Based on our insights into their limitations, we experiment with the recently proposed epsilon-sampling approach, which prunes away all tokens with a probability smaller than epsilon, ensuring that each token in a sample receives a fair probability mass. Through extensive human evaluations, we demonstrate that MBR decoding based on epsilon-sampling significantly outperforms not only beam search decoding, but also MBR decoding with all other tested sampling methods across four language pairs.",
}
| Recent advances in machine translation (MT) have shown that Minimum Bayes Risk (MBR) decoding can be a powerful alternative to beam search decoding, especially when combined with neural-based utility functions. However, the performance of MBR decoding depends heavily on how and how many candidates are sampled from the model. In this paper, we explore how different sampling approaches for generating candidate lists for MBR decoding affect performance. We evaluate popular sampling approaches, such as ancestral, nucleus, and top-k sampling. Based on our insights into their limitations, we experiment with the recently proposed epsilon-sampling approach, which prunes away all tokens with a probability smaller than epsilon, ensuring that each token in a sample receives a fair probability mass. Through extensive human evaluations, we demonstrate that MBR decoding based on epsilon-sampling significantly outperforms not only beam search decoding, but also MBR decoding with all other tested sampling methods across four language pairs. | [
"Freitag, Markus",
"Ghorbani, Behrooz",
"Fern",
"es, Patrick"
] | Epsilon Sampling Rocks: Investigating Sampling Strategies for Minimum Bayes Risk Decoding for Machine Translation | findings-emnlp.617 | 2305.09860 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.618.bib | https://aclanthology.org/2023.findings-emnlp.618/ | @inproceedings{leidinger-etal-2023-language,
title = "The language of prompting: What linguistic properties make a prompt successful?",
author = "Leidinger, Alina and
van Rooij, Robert and
Shutova, Ekaterina",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.618",
doi = "10.18653/v1/2023.findings-emnlp.618",
pages = "9210--9232",
abstract = "The latest generation of LLMs can be prompted to achieve impressive zero-shot or few-shot performance in many NLP tasks. However, since performance is highly sensitive to the choice of prompts, considerable effort has been devoted to crowd-sourcing prompts or designing methods for prompt optimisation. Yet, we still lack a systematic understanding of how linguistic properties of prompts correlate with the task performance. In this work, we investigate how LLMs of different sizes, pre-trained and instruction-tuned, perform on prompts that are semantically equivalent, but vary in linguistic structure. We investigate both grammatical properties such as mood, tense, aspect and modality, as well as lexico-semantic variation through the use of synonyms. Our findings contradict the common assumption that LLMs achieve optimal performance on prompts which reflect language use in pretraining or instruction-tuning data. Prompts transfer poorly between datasets or models, and performance cannot generally be explained by perplexity, word frequency, word sense ambiguity or prompt length. Based on our results, we put forward a proposal for a more robust and comprehensive evaluation standard for prompting research.",
}
| The latest generation of LLMs can be prompted to achieve impressive zero-shot or few-shot performance in many NLP tasks. However, since performance is highly sensitive to the choice of prompts, considerable effort has been devoted to crowd-sourcing prompts or designing methods for prompt optimisation. Yet, we still lack a systematic understanding of how linguistic properties of prompts correlate with the task performance. In this work, we investigate how LLMs of different sizes, pre-trained and instruction-tuned, perform on prompts that are semantically equivalent, but vary in linguistic structure. We investigate both grammatical properties such as mood, tense, aspect and modality, as well as lexico-semantic variation through the use of synonyms. Our findings contradict the common assumption that LLMs achieve optimal performance on prompts which reflect language use in pretraining or instruction-tuning data. Prompts transfer poorly between datasets or models, and performance cannot generally be explained by perplexity, word frequency, word sense ambiguity or prompt length. Based on our results, we put forward a proposal for a more robust and comprehensive evaluation standard for prompting research. | [
"Leidinger, Alina",
"van Rooij, Robert",
"Shutova, Ekaterina"
] | The language of prompting: What linguistic properties make a prompt successful? | findings-emnlp.618 | 2311.01967 | [
"https://github.com/aleidinger/language_of_prompting"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.findings-emnlp.619.bib | https://aclanthology.org/2023.findings-emnlp.619/ | @inproceedings{ravichander-etal-2023-bias,
title = "When and Why Does Bias Mitigation Work?",
author = "Ravichander, Abhilasha and
Stacey, Joe and
Rei, Marek",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.619",
doi = "10.18653/v1/2023.findings-emnlp.619",
pages = "9233--9247",
abstract = "Neural models have been shown to exploit shallow surface features to perform language understanding tasks, rather than learning the deeper language understanding and reasoning skills that practitioners desire. Previous work has developed debiasing techniques to pressure models away from spurious features or artifacts in datasets, with the goal of having models instead learn useful, task-relevant representations. However, what do models actually learn as a result of such debiasing procedures? In this work, we evaluate three model debiasing strategies, and through a set of carefully designed tests we show how debiasing can actually increase the model{'}s reliance on hidden biases, instead of learning robust features that help it solve a task. Further, we demonstrate how even debiasing models against all shallow features in a dataset may still not help models address NLP tasks. As a result, we suggest that debiasing existing models may not be sufficient for many language understanding tasks, and future work should consider new learning paradigms, to address complex challenges such as commonsense reasoning and inference.",
}
| Neural models have been shown to exploit shallow surface features to perform language understanding tasks, rather than learning the deeper language understanding and reasoning skills that practitioners desire. Previous work has developed debiasing techniques to pressure models away from spurious features or artifacts in datasets, with the goal of having models instead learn useful, task-relevant representations. However, what do models actually learn as a result of such debiasing procedures? In this work, we evaluate three model debiasing strategies, and through a set of carefully designed tests we show how debiasing can actually increase the model{'}s reliance on hidden biases, instead of learning robust features that help it solve a task. Further, we demonstrate how even debiasing models against all shallow features in a dataset may still not help models address NLP tasks. As a result, we suggest that debiasing existing models may not be sufficient for many language understanding tasks, and future work should consider new learning paradigms, to address complex challenges such as commonsense reasoning and inference. | [
"Ravich",
"er, Abhilasha",
"Stacey, Joe",
"Rei, Marek"
] | When and Why Does Bias Mitigation Work? | findings-emnlp.619 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.